WO2020250426A1

WO2020250426A1 - Information generation device, corpus production method, and program

Info

Publication number: WO2020250426A1
Application number: PCT/JP2019/023681
Authority: WO
Inventors: 基光白川
Original assignee: ソプラ株式会社
Priority date: 2019-06-14
Filing date: 2019-06-14
Publication date: 2020-12-17
Also published as: JP6635460B1; JPWO2020250426A1

Abstract

[Problem] To automatically generate, from documents belonging to an organization, a corpus for converting organization-internal words to API parameters. [Solution] An information generation device comprising: an intent generation unit that uses a first document to acquire an action name and generates an intent, said first document having the action name or a portion of the action name, said action name being information that specifies a business process, and said intent being information that specifies a business process and being information that has an action name; an entity generation unit that uses a second document to generate an entity name, said second document having an attribute name and said entity name being an attribute name; a processing unit that associates the generated entity name with a variable in an initial template of a conversational sentence having a variable, thereby generating a template having a variable with which the entity name is associated; and an accumulation unit that accumulates intents, entity names, and templates. With this information generation device, a corpus for converting organization-internal words to API parameters can be automatically generated from documents belonging to the organization.

Description

Information generators, corpus production methods, and programs

The present invention relates to an information generator or the like that generates information such as a corpus.

Conventionally, in order to search information in a database in an organization such as a company or an organization, for example, a technology for creating a dedicated program or acquiring information from the database using a search tool and outputting it has been used. It was.

Further, in recent years, there has been a technique for performing an Internet search and turning on / off a predetermined device using words (which may be called conversational sentences) uttered by a user (see, for example, Non-Patent Document 1).

Furthermore, conventionally, a corpus for a computer to process a conversational sentence in natural language to perform a search and a response had to be manually input by a person, for example.

However, conversational sentences (which may be called natural language sentences) change dynamically even for inquiries that request the same search results, depending on, for example, the person who issues them or the organization of the company to which the person belongs. For example, there are many conversational sentences that inquire about the overtime hours of a specific employee in a specific month, such as "Tell me about the overtime hours of Nakamura last month" and "Tell me about the overtime hours of Ichiro Nakamura last month." To do.

With the above-mentioned conventional technology, it was difficult to accept dynamically changing conversational sentences and search for information even with such inquiries requesting the same search results. Therefore, the inventor of the present application has devised an information output device capable of performing such information retrieval by converting words in a conversational sentence into API parameters using an intent, one or more entities, and one or more conversational sentence information. ..

An object of the present invention is to provide an apparatus for automatically generating a corpus for natural language processing of words used in an organization by using one or more documents such as a document file or a database owned by the organization, and in particular, for example, An object of the above information output device is to provide an information generation device capable of automatically generating a corpus containing one or more entity information and one or more conversational sentence information, which is an intent used to convert such words into API parameters. To do.

The information generator of the first invention stores one or more documents including an action name or a first document having a part of the action name and a second document having an attribute name, which are information specifying business processing. An intent generation unit that acquires an action name using the first document stored in the document storage unit and generates an intent that is information that specifies a business process and has an action name. (2) Using the document, the entity generation unit that generates the entity name that is the attribute name and the entity name generated by the entity generation unit are stored in the template storage unit that stores the initial template of the conversation sentence that has the variable. By associating with the variables of the initial template, the processing unit that generates the template with the variables associated with the entity names, the intent generated by the intent generation unit, the entity name generated by the entity generation unit, and the processing unit It is an information generation device including a storage unit for storing the generated template.

With this configuration, intents, entity names, and templates that make up a corpus for natural language processing of words used in the organization can be automatically generated using one or more documents owned by the organization.

In the information generator of the second invention, with respect to the first invention, the second document further has one or more words corresponding to the attribute name, and the entity generation unit is one or more words. The storage unit is an information generation device that further generates the above entities and further stores one or more entities in association with the entity name.

With such a configuration, one or more entities corresponding to the entity name can be automatically generated.

For the second invention, the information generator of the third invention inserts one or more of the one or more entities generated by the entity generator into the variables associated with the entity names in the template. By doing so, a conversation sentence information generation unit that generates one or more conversation sentence information that is information on conversation sentences related to business processing is further provided, and the storage unit is one or more conversation sentence information generated by the conversation sentence information generation unit. Is an information generation device that stores more information in association with intents.

With such a configuration, one or more conversational sentence information corresponding to the intent can be automatically generated. Therefore, using one or more documents possessed by the organization, it is possible to automatically generate intents, entity information, and one or more conversational sentence information that constitute a corpus for converting words used in the organization into API parameters.

In the information generator of the fourth invention, with respect to the third invention, the initial template has one or more variables, and the processing unit uses one or more entity names generated by the entity generation unit as the initial template. Corresponding to one or more variables having, the document storage unit stores one or more table information including an attribute name, a table having one or more words corresponding to the attribute name, and a table identifier that identifies the table. , The first document is an API design document which is a design document having an action name and is a design document related to API, or a design document having a table name or an attribute name including information which can form at least a part of the action name. Yes, it is a DB design document which is a design document related to the database, the second document is each one or more table information stored in the table storage unit, and the intent generation unit acquires the action name from the API design document. Or, the table name or attribute name is acquired from the DB design document, the action name having at least a part of the table name or attribute name is configured, and the intent is the information having the acquired or configured action name. Is generated, and the entity generation unit acquires the attribute name and one or more words corresponding to the attribute name from the table held by the table information for each one or more table information, and the entity name which is the acquired attribute name. , And one or more entities that are one or more acquired words are generated, and the conversational sentence information generation unit acquires one entity from one or more entities generated for each table information of one or more by the entity generation unit. Then, by inserting the acquired one or more entities into one or more variables associated with one or more entity names in the template, one or more conversational sentence information is generated, and the storage unit is intented. One or more conversational sentences generated by the conversational sentence information generation unit, and the intents generated by the generation unit are accumulated, and one or more entities are associated with each one or more entity names generated by the entity generation part. It is an information generation device that stores information in association with the intent.

In the above configuration, a table is a type of database. The table is, for example, a tabular database. The tabular database is, for example, a relational database table, a file such as Excel, but the type does not matter. However, the table does not have to have a tabular format in advance, and may be, for example, a database such as a document file that has been converted into a tabular format or can be converted into a tabular format. The document file that can be converted to the tabular format is, for example, a word file, but the type does not matter. In the case of a document file such as a word, for example, by providing a delimiter in the document, conversion to a tabular format is possible. Further, the table may be of any type as long as it is a database in tabular form, converted to tabular form, or convertible to tabular form.

In the above configuration, one or more of the document storage unit, the table storage unit, and the template storage unit may exist outside the information generation device. It should be noted that such a matter also applies to an intent storage unit, an entity storage unit, a conversation sentence information storage unit, an entity mapping information storage unit, a PK item storage unit, and the like.

With this configuration, API design document or DB design document which is a kind of first document, and table information which is a kind of second document, intent which is information for converting words into API parameters, one or more entities. Information and one or more conversational text information can be automatically generated.

Further, the information generation device of the fifth invention further includes an intent storage unit for storing one or more intents with respect to the fourth invention, and the intent generation unit is described from the API design document. The API information extraction means for extracting one or more API information which is information having an action name and is information about the API, the action name acquisition means for acquiring the action name from each one or more API information, and the action name acquisition means It is provided with an intent generation means for generating an intent having one or more acquired action names, and the storage unit stores one or more intents generated by the intent generation means in the intent storage unit. It is an information generation device provided with storage means.

With this configuration, intents can be automatically generated from the API design document.

Further, the information generation device of the sixth invention further includes an intent storage unit for storing one or more intents with respect to the fourth invention, and the intent generation unit is 1 from the DB design document. At least a part of the table name or attribute name is extracted and extracted for each of the above table names or one or more attribute names and one or more table names or one or more attribute names. It includes an action name generating means for generating an action name having a part and an intent generating means for generating an intent having one or more action names generated by the action name generating means, and the storage unit generates an intent. It is an information generation device including an intent storage means for accumulating one or more intents generated by the means in an intent storage unit.

With this configuration, intents can be automatically generated from the DB design document.

Further, the information generation device of the seventh invention further includes an entity storage unit for storing one or more entities for the fifth or sixth invention, and the entity generation unit is one or more tables. For each information, the entity name generating means that acquires the attribute name from the table information and generates the entity name that is the attribute name, and the table corresponding to the entity name for each one or more entity names generated by the entity name generating means. It is provided with an entity generation means for acquiring one or more words from and generating one or more entities which are one or more words, and the storage unit corresponds to each one or more entity names and corresponds to the entity name. It is an information generation device further comprising an entity storage means for storing one or more entities, which are one or more words acquired from a table, in an entity storage unit.

With such a configuration, an entity name and entity information that is a set of one or more entities corresponding to the entity name can be automatically generated from each of one or more tables.

Further, the information generation device of the eighth invention further includes a conversational sentence information storage unit for storing one or more conversational sentence information with respect to the seventh invention, and the template storage unit has an entity name. It is a template having one or more associated variables, two or more templates having different positions or numbers of variables are stored, and the conversation sentence information generation unit has one or more for each of the two or more templates. An entity acquisition means for acquiring one entity corresponding to the entity name associated with each variable of the above from the entity storage unit, and an entity at the position of one or more variables of the template for each of two or more templates. It is provided with a conversation sentence generation means that generates one or more conversation sentence information by inserting an entity acquired by the acquisition means, and the storage unit inserts one or more conversation sentence information generated by the conversation sentence generation means. It is an information generation device further provided with a conversational sentence information storage means that is associated with a tent and stored in a conversational sentence information storage unit.

With such a configuration, it is a template having one or more variables associated with entity names, and various conversational sentence information can be automatically generated by using two or more templates having different positions or numbers of variables.

Further, the information generation device of the ninth invention has an entity mapping information storage unit for storing one or more entity mapping information and one or more table information for each of the seventh or eighth invention. From the table information, obtain the table identifier, the translation item name that is the attribute name in the table identified by the table identifier and is the information that identifies the entity to be converted, and one entity corresponding to the translation item. In addition, the storage unit further includes an entity mapping information generation unit that acquires an entity name corresponding to one entity from the entity storage unit and generates entity mapping information having a table identifier, a translation item name, and an entity name. , An information generation device further comprising an entity mapping information storage unit that stores one or more entity mapping information generated by the entity mapping information generation unit in the entity mapping information storage unit.

With such a configuration, one or more entity mapping information having a table identifier, a translation item name, and an entity name can be automatically generated from one or more table information.

Further, the information generator of the tenth invention has a PK item storage unit for storing one or more PK items and one or more table information for each of the fourth to ninth inventions. , PK item generator that acquires a table identifier that identifies a table and a primary key identifier that identifies a primary key of a table from the table information, and generates a PK item that is information having the table identifier and the primary key identifier. The storage unit is an information generation device further comprising a PK item storage unit that stores one or more PK items generated by the PK item generation unit in the PK item storage unit.

With such a configuration, one or more PK items having a table identifier and a primary key identifier can be automatically generated from one or more table information.

According to the present invention, it is possible to automatically generate a corpus for natural language processing of words used in an organization by using one or more documents such as a document file or a database owned by the organization. In addition, it is a conversational sentence used in the organization, and it is a corpus used by an information output device that accepts a conversational sentence that changes dynamically even in a query for the same search result and searches for information, and is used to convert words into API parameters. An information generator is realized that automatically generates a corpus including intents that are information, one or more entity information, and one or more conversational sentence information.

Block diagram of the information output device according to the first embodiment Flow chart explaining the operation of the information output device Flow chart explaining the parameter acquisition process Data structure diagram of the table information Data structure diagram of the intent The figure which shows an example of the contents of the API information storage part The figure which shows the figure which shows an example of the contents of the conversation sentence information storage part The figure which shows the example of one or more entities corresponding to each entity name. Data structure diagram of the same entity mapping information Data structure diagram of the same PK item External view of the computer system Diagram showing an example of the internal configuration of the computer system Block diagram of the information generator according to the second embodiment The figure which shows an example of the contents of the template storage part Flow chart explaining the operation of the information generator A flowchart illustrating a detailed operation example of the information generator. Diagram showing an example of the document file Data structure diagram of the intent name dictionary Data structure diagram of the same entity name dictionary Block diagram of the information generator according to the third embodiment Block diagram of the information processing device according to the fourth embodiment Block diagram of the information processing device according to the fifth embodiment

Hereinafter, embodiments of the information generation device, information output device, etc. will be described with reference to the drawings. Since the components with the same reference numerals perform the same operation in each embodiment, the description may be omitted again.

In the first embodiment, an information output device that is a conversational sentence used in an organization and can accept a conversational sentence that dynamically changes even in an inquiry for the same search result and perform an information search will be described.

In the second embodiment, the corpus used by the information output device is an intent for converting words used in the organization into API parameters, and a corpus including one or more entity information and one or more conversational sentence information. , A device that automatically generates using one or more documents possessed by an organization will be described. In particular, an information generator that automatically generates such a corpus using an API design document, which is a kind of first document, and table information, which is a kind of second document, will be described.

In the third embodiment, the DB design document, which is a kind of the first document, and the information generation device that automatically generates the corpus using the table information will be described.

In the fourth embodiment, the API design document and the table information are used to automatically generate a corpus containing intent, one or more entity information, and one or more conversational sentence information, and the corpus is used in the organization. An information processing device that is a conversational sentence to be used and can accept a conversational sentence that changes dynamically even in a query for the same search result and search for information will be described.

In the fifth embodiment, an information processing device capable of automatically generating the corpus using the DB design document and the table information, accepting the conversational sentence, and searching for information using the corpus will be described.

(Embodiment 1)
The information output device 500 in this embodiment is, for example, a server. The server is, for example, a server in a company that provides services such as information retrieval, but may be a cloud server, an ASP server, or the like, and its type does not matter. The server is communicably connected to one or more terminals (not shown) via a network such as a LAN or the Internet, a wireless or wired communication line, or the like. The terminal is, for example, a mobile terminal of a user who uses a service such as information retrieval, but may be a stationary PC or the like, and the type is not limited. The mobile terminal is, for example, a tablet terminal, a smartphone, a mobile phone, a notebook PC, or the like, but the type is not limited.

However, the information output device 500 may be, for example, a stand-alone terminal, and the means for realizing it does not matter.

FIG. 1 is a block diagram of the information output device 500 according to the present embodiment. The information output device 500 includes a storage unit 1, a reception unit 2, a processing unit 3, and an output unit 4. The storage unit 1 includes a table storage unit 11, an intent storage unit 12, an API information storage unit 13, a conversation text information storage unit 14, an entity storage unit 15, an entity mapping information storage unit 16, a PK item storage unit 17, and a day conversion. The information storage unit 18 is provided. The reception unit 2 includes a conversation text reception unit 21. The conversation sentence reception unit 21 includes a voice reception means 211 and a voice recognition means 212.

The processing unit 3 includes an intent determination unit 31, a conversation sentence information determination unit 32, an entity acquisition unit 33, a parameter acquisition unit 34, an API information acquisition unit 35, an inquiry information configuration unit 36, and a search unit 37. The parameter acquisition unit 34 includes a determination unit 341, a day information acquisition unit 342, an entity name acquisition unit 343, a translation item name acquisition unit 344, a table identifier acquisition unit 345, a primary key identifier acquisition unit 346, and a conversion parameter acquisition unit 347. .. The output unit 4 includes a search result output unit 41.

The storage unit 1 can store various types of information. The various types of information include, for example, tables, intents, API information, corpora, entities, entity mapping information, PK items, and day conversion information. Information on tables and the like will be described later. In addition, other information will be explained in a timely manner.

One or two or more tables are stored in the table storage unit 11. A table is a type of database used for business processing. It can be said that a database is a set of information structured so that a computer can easily search and store it.

Business processing is processing related to business. It can be said that the work is, for example, a work performed by an organization such as a company or an organization. The work includes, for example, work until the fixed time, overtime after the fixed time, vacation, absenteeism, A case, B case, accounting, manufacturing, etc., but the type does not matter. The business process is, for example, a search for information. The search may be referred to as an inquiry, for example. The information to be searched is, for example, attendance information such as overtime hours and working hours, but may be financial information such as balances and stock prices, or operating conditions of manufacturing lines, and the contents thereof are not limited.

However, the business processing may be, for example, registration of information or processing based on information (for example, payroll calculation based on attendance information, financial transaction based on financial information, etc.), and the type does not matter. Further, it may be said that the database used for business processing is, for example, a local database in the organization.

The table may be, for example, a tabular database. For example, one or more item names are registered in the table, and one or two or more values are registered for each of the one or more item names. The item name may be referred to as an attribute name, and each value of 1 or more corresponding to one item name may be referred to as an attribute value. The table is, for example, a relational database table, TSV, Excel, CSV, or the like, but the type is not limited. It can be said that the attribute name is the name of the attribute and the attribute value is a value that the attribute can take. An attribute is a property of data and may be an item of a field in a relational database.

However, the table does not have to have a tabular format in advance, and may be, for example, a database of document files that have been converted to a tabular format or can be converted to a tabular format. The document file that can be converted to the tabular format is, for example, a word file, but it may be a file such as an email, chat, or text, and the type is not limited. In the case of a document file such as a word, conversion to a tabular format is possible by providing a delimiter in the character string, for example. Further, the table is a database used for business processing, and any type of database can be used as long as it is a database in tabular form, converted to tabular form, or convertible to tabular form.

Specifically, the stored table is, for example, an employee master, a business master, a working time table, or the like. Master data about employees is registered in the employee master. The master data related to an employee is, for example, data such as an employee code and an employee name, but the type does not matter. Master data related to the business is registered in the business master. The master data related to the business is, for example, data such as a business ID and a business name, but the type thereof does not matter.

Data related to working hours are registered in the working hours table. The data related to working hours is, for example, variable data such as date, overtime hours, total working hours, business ID, etc., but the type is not limited. The total working hours is the total of working hours up to the scheduled time and overtime hours.

Normally, data related to working hours as described above is registered in the working hours table in association with the employee code. However, the information to which the data on working hours is associated may be, for example, an employee name, or any information that can identify the employee.

One or two or more intents are stored in the intent storage unit 12. An intent is information managed for each business process. The management for each business process may be, for example, stored in the intent storage unit 12 in association with the information that identifies the business process. The information that identifies the business process is usually an action name described later, but the format does not matter. Correspondence also includes, for example, the case where the intent has information that identifies the business process.

The intent usually has an action name that identifies the business process. The action name is the name of the action. An action is a business process that is usually executed via an API. However, the action may be, for example, a business process executed in response to an SQL statement.

The action name usually corresponds to the API information described later. Therefore, it may be considered that the intent is associated with the API information, for example, via the action name.

Intents also usually have an intent name. The intent name is the name of the intent. Therefore, the intent can be considered as information that associates the intent specified by the intent name or the intent name with the action name or the action specified by the action name.

Specifically, the action name is, for example, "overtime hour inquiry", "working hour data inquiry", etc., but any information that can identify the business process may be used.

One or two or more API information is stored in the API information storage unit 13. API information is information about API. It can be said that the API is an application programming interface and is an interface for using the functions of the program. An API is software such as a function, method, or execution module, for example. The API is, for example, WebAPI, but other APIs may be used. A Web API is an API constructed by using a Web communication protocol such as HTTP or HTTPS. Since APIs such as WebAPI are known techniques, detailed description thereof will be omitted.

API information is information that corresponds to the intent. As described above, the API information corresponds to the intent, for example, through the action name.

API information is usually information for performing information retrieval. However, the API information may be, for example, information for registering information or performing processing based on the information.

API information has one or more parameter specific information. The parameter specific information is information for specifying a parameter. It may be said that the parameter is a value having a specific attribute. The value is usually a variable. Variables may be called arguments.

The parameter is usually the information obtained by converting the entity, but it may be the entity itself. It can be said that an entity is a word included in a conversational sentence. The entity will be described later. The parameters are, for example, arguments given to the API or variables in the SQL statement.

The parameter may be composed of, for example, a set of an attribute name and a value. Specifically, the set of the attribute name and the value is, for example, "shain_code = 2", "sta_date = 2019041, end_date = 20190430", but the format does not matter.

The parameter specific information is, for example, a parameter name. The parameter name is the name of the parameter. Alternatively, the parameter-specific information is, for example, an attribute name, but any information that can specify the parameter may be used.

API information is, for example, a method. The methods are, for example, "search_overtime hours (employee code = $ 1, start date = $ 2, end date = $ 3)" and "search_business hours (employee code = $ 1, work date = $ 2, business ID =)". $ 3) "etc., but the format does not matter. Since the method is known, detailed description thereof will be omitted.

Alternatively, the API information may be, for example, an SQL statement. It can be said that the SQL statement is a statement for making an inquiry to the database. The SQL statement is, for example, "select_overtime_from_table name = $ 1_where_employee code = $ 2, business date = $ 3", but the structure does not matter. Since the SQL statement is well known, detailed description thereof will be omitted.

Alternatively, the API information may be, for example, a set of a URL and a method, and its structure does not matter. Specifically, the set of the URL and the method is, for example, "http://develop.soppra ...? Employee code = $ 1 & start date = $ 2 & end date = $ 3". Does not matter. Note that this type of API information may be referred to as Web API information, for example.

Conversational sentence information storage unit 14 stores one or more conversational sentence information. Conversational sentence information is information on conversational sentences. The conversational sentence information is usually an example sentence of a conversational sentence. Example sentences are, for example, "Tell me about Nakamura's overtime hours last month" and "Tell me about Nakamura's working hours for yesterday's A project", but the content does not matter.

However, the conversational sentence information may be a template of the conversational sentence. The template is, for example, "Tell me {Overtime entity} of {Date entity} of {Employee entity}" or "Tell me {Working time entity} of {Business entity} of {Date entity} of {Employee entity}". Etc., but the format does not matter. Information expressed by "{", "}" such as {employee entity} included in the template may be called a variable, for example. Further, the entity and the entity name such as "employee entity" will be described later.

Conversational text information usually corresponds to the intent. That is, for example, one or more conversational sentence information is stored in the conversational sentence information storage unit 14 for each one or more intents stored in the intent storage unit 12 (see FIG. 7: described later). ).

In addition, the conversation sentence information storage unit 14 usually stores one or more entity information for each one or more stored conversation sentence information. Entity information is information about an entity. The entity information is, for example, information about one or more entities corresponding to one conversational sentence information. The entity information may be, for example, an entity, an entity name, or information indicating the position or order of the entity in a conversational sentence. The entity information has, for example, an entity and an entity name. The entity information has, for example, an entity, a start position, an end position, and an entity name when the conversational sentence information is an example sentence.

The entity possessed by the entity information is a representative value among one or two or more entities corresponding to the entity name possessed by the entity information. The representative value is, for example, the first entity in one or more entities corresponding to one entity name, but any entity may be used.

The start position is the position where the entity starts in the conversation text information. The start position is represented by, for example, a value (for example, "1", "4", etc.) indicating which character the first character of the entity is in the character string constituting the conversation sentence. Similarly, the end position is a position where the entity ends in the conversation text information, and is, for example, a value indicating the number of the last character of the entity (for example, "2", "5", etc.). ). However, the expression format of the start position and the end position does not matter. The start position and end position may be referred to as offsets. Further, the offset may be expressed by the number of bytes, and its format does not matter.

The entity name is the name of the entity. The name of the entity may be, for example, the attribute name of the entity. The entity name is, for example, "employee entity", "date entity", "overtime hour entity", etc., but the format is not limited as long as it is information that can express the attributes of the entity. An employee entity is an entity related to an employee. A date entity is an entity related to a date. The overtime hours entity is an entity related to overtime hours.

Alternatively, the entity information may have, for example, an entity name, an entity, and an order information when the conversational sentence information is a template. The order information is a value indicating which variable the entity name corresponds to in one or more variables included in the template. However, the entity information may be, for example, an entity name and a set of one or two or more entities corresponding to the entity name, and the structure thereof does not matter.

Specifically, in the conversation text information storage unit 14, for example, conversation text information "Tell me about Nakamura's overtime hours last month" and three corresponding entity information "Nakamura, 1, 2, employee entities" , "Last month, 4, 5, date entity", and "overtime hours, 7, 10, overtime entity" may be stored.

The corpus in the present embodiment is a set of information for the computer to perform natural language processing, and is a set of information related to conversational sentences used in the organization. The corpus may be considered as each of one or more conversational sentence information stored in the conversational sentence information storage unit 14, for example, one or more conversational sentence information, and entity information corresponding to each conversational sentence information. Can be thought of as a set of. Alternatively, in addition to one or more conversational sentence information, the corpus also includes, for example, an intent for generating the one or more conversational sentence information, an entity name, and one or more entities corresponding to the entity name. It may be considered to include.

One or two or more entities are stored in the entity storage unit 15. An entity is one or more words associated with conversational sentence information. A word is a unit that makes up a sentence. A word may be simply called, for example, a "word" or a "word", or may be considered as a kind of morpheme (for example, an independent word described later).

In the entity storage unit 15, for example, one or two or more entities are stored in association with each one or more entity names.

The entity usually corresponds to any one or more of the conversational text information stored in the conversational text information storage unit 14. Therefore, for example, one or more entities may be stored in the entity storage unit 15 for each one or more conversational sentence information stored in the conversational sentence information storage unit 14.

The word corresponding to the conversational sentence information is usually an independent word. An independent word is a word that can form a phrase by itself, and is, for example, a noun, but may also be a verb, an adjective, or the like. However, the corresponding word may be an attached word. An adjunct word is a word that cannot form a phrase by itself and that forms a phrase with other independent words, such as auxiliary verbs and particles. That is, the corresponding word is usually an independent word, but may be an independent word with an adjunct.

Also, the corresponding words may be, for example, collocations. A collocation is a word that expresses a certain meaning by connecting two or more autonomous words, and may be called a compound word. The collocations are, for example, "overtime hours" in which "overtime" and "hours" are combined, "A project" in which "A" and "project" are combined, etc., but a group named "Ichiro Nakamura". However, any set of two or more words may be used.

When the conversational sentence information is an example sentence, the corresponding words are one or more words (that is, the information in the conversational sentence information) included in the example sentence, but when the conversational sentence information is a template. Is a word (that is, information added to conversational sentence information) to be inserted at the position of one or more variables included in the template.

One or two or more entity mapping information is stored in the entity mapping information storage unit 16. It can be said that the entity mapping information is information that associates the table identifier and the translated item name with the entity name.

The entity mapping information has, for example, a table identifier, a translation item name, and an entity name. The table identifier is information for identifying one or more tables stored in the table storage unit 11. The table identifier is, for example, "MST_SHAIN" or "TRN_TIME_CARD", but any information that can identify the table may be used.

The translation item name is an attribute name in the table identified by the table identifier, and is information that identifies the entity to be translated. The converted entity is the entity converted to the parameter (for example, the entity "Ichiro Nakamura" converted to the parameter "shain_code = 1" and the entity "last month" converted to the parameter "sta_date = 2019041, end_date = 20190430". "Etc.).

The translated item name is, for example, an item name corresponding to a key other than the primary key among one or more item names of the table identified by the table identifier, and the item name to be converted into the item name corresponding to the primary key. It may be. The item name corresponding to the primary key is, for example, a primary key identifier.

The translation item name is, for example, "SHAIN_CODE" or "TIME_CARD_DATE", but any information that can identify the entity to be converted may be used.

One or two or more PK items are stored in the PK item storage unit 17. The PK item is an item corresponding to the primary key. The PK item has, for example, a table identifier and a primary key identifier. The primary key identifier is information that identifies the primary key of the table identified by the table identifier.

The primary key identifier is, for example, "SHAIN_CODE" in the case of a table identified by the table identifier "MST_SHAIN" (hereinafter, employee master), and is a table (hereinafter, working hours table) identified by the table identifier "TRN_TIME_CARD". There are two cases, "SHAIN_CODE" and "TIME_CARD_DATE".

One or more day conversion information is stored in the day conversion information storage unit 18. The day conversion information is information for converting a day word into a date. A day word is a word about a day. The day word is usually a word associated with the entity name "date entity", for example, "last month", "yesterday", "last week", "this year", "this month", "last year", "previous term", For example, "this year", but any information that can be converted into a date may be used.

The day conversion information includes a day word and day information acquisition information. The daily information acquisition information is information for acquiring the daily information. The day information is information about the day corresponding to the day word, and is information used when constructing inquiry information. The day information may be, for example, information indicating a date such as "April 1" or information indicating a period from a start date to an end date such as "4/1 to 4/30", and the format thereof is It doesn't matter. The day information acquisition information is, for example, a function name or a method name, but may be API information or the program itself, and its format does not matter.

Specifically, the day information acquisition information for the day word "last month" is, for example, the current time information (for example, "May 10, 2019 11:15": the same applies hereinafter), and the current time information has the information. The previous month (for example, "April") is acquired for the month (for example, "May"), and the day information (for example, the first day to the last day of the previous month) is referred to by referring to the calendar information of the previous month. For example, a program or the like for acquiring "4/1 to 4/30" etc. may be used.

Further, the day information acquisition information for the day word "this year" is, for example, from the first day of the year by acquiring the current time information and referring to the calendar information of the year (for example, "2019") that the current time information has. , API information or the like for acquiring day information (for example, “2019/1/1 to 2019/5/10”) up to the date of the current time information.

Further, the day information acquisition information for the day word "yesterday" is a method of acquiring the current time information and acquiring the day information of the day before the day held by the current time information (for example, "5/9"), or The method name or the like may be used.

Reception department 2 receives various information. The various types of information are, for example, conversational sentences. The reception unit 2 receives information such as conversational sentences from, for example, a terminal, but may receive information via an input device such as a keyboard, a touch panel, or a microphone. Alternatively, the reception unit 2 may accept information read from a recording medium such as a disk or a semiconductor memory, and the mode of acceptance is not limited.

The conversation sentence reception unit 21 accepts conversation sentences. A conversational sentence is a sentence in which a person speaks, and can be said to be a sentence in natural language. Conversational sentences are accepted, for example, by voice, but may be accepted by text. A voice is a voice made by a person. A text is a character string that voice-recognizes a voice uttered by a person. A character string is composed of an array of one or more characters.

The voice receiving means 211 receives the voice of the conversational sentence. The voice receiving means 211 receives the voice of the conversational sentence from the terminal, for example, in pairs with the terminal identifier, but may receive the voice through the microphone. The terminal identifier is information that identifies a terminal. The terminal identifier is, for example, a MAC address, an IP address, an ID, or the like, but any information that can identify the terminal may be used. The terminal identifier may be a user identifier that identifies the user of the terminal. The user identifier is, for example, an e-mail address, a telephone number, or the like, but may be an ID, an address, a name, or the like, and may be any information that can identify the user.

The voice recognition means 212 performs voice recognition processing on the voice received by the voice reception means 211, and acquires a conversational sentence which is a character string. The voice recognition process is a known technique, and detailed description thereof will be omitted.

The processing unit 3 performs various processes. The various processes include, for example, an intent determination unit 31, a conversation sentence information determination unit 32, an entity acquisition unit 33, a parameter acquisition unit 34, an API information acquisition unit 35, an inquiry information configuration unit 36, a search unit 37, and a determination means. 341, day information acquisition means 342, entity name acquisition means 343, translation item name acquisition means 344, table identifier acquisition means 345, primary key identifier acquisition means 346, conversion parameter acquisition means 347, and the like. Further, the various processes include, for example, various discriminations described in the flowchart.

The processing unit 3 processes, for example, the intent determination unit 31 and the like in response to the conversation sentence reception unit 21 receiving the conversation sentence. When a conversation sentence is transmitted from one or more terminals to a pair with the terminal identifier, the processing unit 3 performs processing of the intent determination unit 31 and the like for each one or more terminal identifiers.

The intent determination unit 31 determines the intent corresponding to the conversation sentence received by the conversation sentence reception unit 21.

Specifically, the intent determination unit 31 first acquires, for example, the text corresponding to the conversation sentence received by the conversation sentence reception unit 21. As described above, the text is, for example, the result of voice recognition of the conversation sentence received by the conversation sentence reception unit 21, but may be the conversation sentence itself received by the conversation sentence reception unit 21.

That is, when the intent determination unit 31 receives a conversational sentence that is voice, the intent determination unit 31 voice-recognizes the conversational sentence and acquires the text. When the conversational sentence which is a text is accepted, the intent determination unit 31 may acquire the text.

Next, the intent determination unit 31 acquires one or more independent words from the acquired text by, for example, performing morphological analysis on the acquired text. The morphological analysis is a known technique, and detailed description thereof will be omitted.

Then, the intent determination unit 31 determines an intent having an action name having a word that is the same as or similar to the acquired one or more independent words.

For details, for example, a synonym dictionary is stored in the storage unit 1. A synonym dictionary is a dictionary related to synonyms. In the synonym dictionary, a word possessed by the action name and one or more synonyms of the word are registered for each action name constituting one or more intents stored in the intent storage unit 12. ing. Specifically, for example, "overtime", "overtime hours", and the like may be registered in association with the word "overtime hours" of the action name "overtime hours inquiry".

For example, when the conversation sentence reception unit 21 accepts the conversation sentence "Tell me about the overtime hours of Otsuka last month", the intent determination unit 31 will use the conversation sentence "Otsuka", "last month", or "overtime hours". Etc. is acquired, and the intent storage unit 12 is searched using each independent word as a key, and it is determined whether or not there is an intent having an action name that matches the independent word. The match is, for example, an exact match, but may be a partial match. Then, when there is an intent having an action name having a word matching the independent word, the intent determination unit 31 determines the intent. In this example, since there is an intent having the action name "overtime hours reference" having the word "overtime hours" matching the independent word "overtime hours", the intent is determined.

If there is no intent having an action name having a word that matches the independent word, for example, the intent determination unit 31 refers to one synonym from one or more synonyms corresponding to the independent word from the synonym dictionary. Is acquired, the intent storage unit 12 is searched using the one synonym as a key, and it is determined whether or not there is an intent having an action name having a word matching the one synonym. Then, when there is an intent having an action name having a word matching the one synonym, the intent determination unit 31 determines the intent. If there is no such intent, the intent determination unit 31 performs the same processing for other synonyms to determine the intent. For any synonym, if there is no such intent, the intent determination unit 31 may output that the intent is not determined.

The conversation sentence information determination unit 32 searches the conversation sentence information storage unit 14 using the intent determined by the intent determination unit 31 as a key, and receives the conversation sentence from one or more conversation sentence information corresponding to the intent. The conversation sentence information most similar to the conversation sentence received by the part 21 is determined.

The conversational sentence information that most closely resembles the conversational sentence is, for example, the conversational sentence information having the highest degree of similarity to the conversational sentence. That is, the conversation sentence information determination unit 32 calculates, for example, the degree of similarity between the received conversation sentence and one or more conversation sentence information corresponding to the determined intent, and the conversation sentence information having the maximum similarity degree. To decide.

Alternatively, the conversation sentence information determination unit 32 may search for a conversation template that matches the template in which the position of the noun of the accepted conversation sentence is used as a variable. That is, the conversation sentence information storage unit 14 stores a template in which one or two or more entity names are used as variables, and the conversation sentence information determination unit 32 stores one or two or more of the received conversation sentences. The position of the entity name is acquired, and the template corresponding to the position of the acquired entity name is determined as the conversational sentence information. The position of one or more entity names in the conversation sentence is information indicating the number of the entity name in the template having one or more entity names.

The entity acquisition unit 33 corresponds to one or more entities corresponding to the conversation sentence information determined by the conversation sentence information determination unit 32, and is one or more words possessed by the conversation sentence received by the conversation sentence reception unit 21. Get an entity.

For example, the entity acquisition unit 33 acquires the start position and the end position of each one or more entities corresponding to the determined conversation sentence information from the conversation sentence information storage unit 14, and receives the conversation. Obtain the word specified by the start position and the end position from the sentence.

Specifically, as described above, for example, the conversation sentence information "Tell me the overtime hours of Nakamura last month" and the corresponding three entity information "Nakamura, 1, 2" are sent to the conversation sentence information storage unit 14. , Employee entity "," Last month, 4, 5, Date entity ", and" Overtime hours, 7, 10, Overtime entity "are stored, and the conversation sentence reception department 21 stores the conversation sentence" Otsuka's last month's overtime hours. When "Tell me" is accepted and the conversation sentence information determination unit 32 determines the above conversation sentence information "Tell me the overtime work hours of Nakamura last month", the entity acquisition unit 33 receives the conversation sentence "Otsuka's overtime work last month". From "Tell me the time", the word "Otsuka" specified by the start position "1" and the end position "2" of the first entity information, the start position "4" of the second entity information, and The word "last month" specified by the end position "5" and the word "overtime hours" specified by the start position "7" and the end position "10" of the third entity information are acquired.

The parameter acquisition unit 34 acquires one or more parameters corresponding to one or more entities acquired by the entity acquisition unit 33.

The acquired parameter is, for example, the acquired entity itself, but it may be information obtained by converting the acquired entity. That is, for example, when the acquired day word is included in one or more entities, the parameter acquisition unit 34 converts the day word into the parameter day information.

The determination means 341 constituting the parameter acquisition unit 34 determines whether or not the day word exists in one or more entities acquired by the entity acquisition unit 33. Specifically, for example, one or two or more day words are stored in the storage unit 1, and the determination means 341 matches any of the stored day words for each of the acquired one or more entities. It is determined whether or not to do so, and when the determination results for at least one entity show a match, it is determined that the day word exists in the acquired one or more entities.

When the determination means 341 determines that the day word exists in one or more acquired entities, the day information acquisition means 342 acquires the day conversion information corresponding to the day word from the day conversion information storage unit 18. , The day information which is a parameter is acquired by using the day conversion information.

Specifically, for example, the day word "last month" and the like are stored in the storage unit 1, and the conversational sentence "Tell me the overtime hours of last month of Otsuka" is accepted, and the three entities "Otsuka" and " When "last month" and "overtime hours" are acquired, the determination means 341 determines that the day word exists among the acquired 3 entities because the entity "last month" matches the day word "last month". To do. For example, the current time information is acquired and the day information (for example, "4/1 to 4/30") is acquired.

The day information acquisition means 342 acquires the day information acquisition information (for example, a program) corresponding to the day word "last month" from the day conversion information storage unit 18. Then, the day information acquisition means 342 acquires the current time information (for example, "May 10, 2019 11:15") from the built-in clock of the MPU, the NTP server, or the like by using the day information acquisition information, and the relevant day information acquisition means 342. Acquires the previous month (for example, "April") with respect to the month (for example, "May") that the current time information has. Then, the day information acquisition means 342 acquires the day information "April 1st to April 30th" from the first day to the last day of the previous month by referring to the calendar information of the previous month.

When the day word acquired from the conversation sentence is "this year", the day information acquisition means 342 stores the day information acquisition information (for example, API information) corresponding to the day word "this year" in the day conversion information storage unit. Obtained from 18. Then, the day information acquisition means 342 acquires the current time information from the built-in clock or the like by using the day information acquisition information, and refers to the calendar information of the year (for example, "2019") possessed by the current time information. , The day information (for example, "2019/1/1 to 2019/5/10") from the first day of the year to the day held by the current time information is acquired.

Further, when the acquired day word is "yesterday", the day information acquisition means 342 acquires the day information acquisition information (for example, a method) corresponding to the day word "yesterday" from the day conversion information storage unit 18. .. Then, the day information acquisition means 342 acquires the current time information from the built-in clock or the like by using the day information acquisition information, and the day information of the day before the day that the current time information has (for example, "5/9"). ”) Is acquired.

The entity name acquisition means 343 acquires the entity name corresponding to the entity from the entity storage unit 15 for each one or more entities acquired by the entity acquisition unit 33.

The entity name corresponding to the entity is an entity name paired with a start position and an end position that match or are similar to the position of the entity corresponding to the entity in the conversation sentence obtained by the entity. The entity name acquisition means 343 acquires the entity name corresponding to the entity from the entity storage unit 15 for each one or more entities acquired by the entity acquisition unit 33, for example, using the entity information associated with the entity. You may.

Specifically, for example, when the three entities "Otsuka", "Last month", and "Overtime hours" are acquired from the accepted conversation sentence "Tell me the overtime hours of Otsuka last month", the entity name acquisition means. 343 accepts among the three entity information stored in association with the conversational sentence information in the conversational sentence information "Tell me Nakamura's overtime hours last month" stored in the conversational sentence information storage unit 14. Corresponds to "Nakamura" using the first entity information that has the same start position "1" and end position "2" as "Otsuka" in the conversation sentence "Tell me about the overtime hours of Otsuka last month". To get the "employee entity".

Further, for example, the entity name acquisition means 343 sets the same start position "4" and end position "5" as "last month" in the conversation sentence "Tell me the overtime hours of last month of Otsuka" among the above three entity information. Use the second entity information you have to get the "date entity" that corresponds to "last month", and the same start position as "overtime hours" in the conversation "Tell me about overtime hours last month in Otsuka". Using the third entity information having "7" and the end position "10", the "overtime hours entity" corresponding to the "overtime hours" is acquired.

The translation item name acquisition means 344 acquires, for example, a translation item name paired with the entity name from the entity mapping information storage unit 16 for each one or more entity names acquired by the entity name acquisition means 343.

Specifically, the entity mapping information storage unit 16 has, for example, an entity mapping information 1 having a table identifier "MST_SHAIN", a translation item name "SHAIN_NAME", and an entity name "employee entity", a table identifier "TRN_TIME_CARD" and a translation item. Entity mapping information 2 having the name "TIME_CARD_DATE" and the entity name "date entity", entity mapping information 3 having the table identifier "TRN_TIME_CARD", the translation item name "ZIKANGAI_TIME" and the entity name "overtime entity" are stored. When the entity name acquisition means 343 acquires the three entity names "employee entity", "date entity", and "overtime entity", the translation item name acquisition means 344 uses the three entity mapping informations 1 to 3. Translated item name "SHAIN_NAME" paired with the entity name "employee entity", translated item name "TIME_CARD_DATE" paired with the entity name "date entity", and translated item name paired with "overtime entity" Acquire "ZIKANGAI_TIM".

The table identifier acquisition means 345 acquires a table identifier paired with the entity name from the entity mapping information storage unit 16 for each one or more entity names acquired by the entity name acquisition means 343.

Specifically, the table identifier acquisition means 345 uses, for example, the above three entity mapping information 1 to 3 to form a table identifier “MST_SHAIN” paired with the entity name “employee entity” and an entity name “date entity”. Acquire the paired table identifier "TRN_TIME_CARD" and the paired table identifier "TRN_TIME_CARD" with the entity name "overtime entity".

The primary key identifier acquisition means 346 acquires a primary key identifier paired with the table identifier acquired by the table identifier acquisition means 345 from the PK item storage unit 17.

Specifically, the PK item storage unit 17 has, for example, a PK item 1 having a table identifier "MST_SHAIN" and a primary key identifier "SHAIN_CODE", and a PK item having a table identifier "TRN_TIME_CARD" and a primary key identifier "SHAIN_CODE". 2. PK item 3 having the table identifier "TRN_TIME_CARD" and the primary key identifier "TIME_CARD_DATE", PK item 4 having the table identifier "MST_GYOUMU" and the primary key identifier "GYOUMU_ID", etc. are stored, and the primary key identifier Using the PK items 1 to 4, the acquisition means 346 uses the primary key identifier "SHAIN_CODE" paired with the table identifier "MST_SHAIN", the primary key identifier "SHAIN_CODE" paired with the table identifier "TRN_TIME_CARD", and the table identifier " Acquires the primary key identifier "TIME_CARD_DATE" paired with "TRN_TIME_CARD" and the primary key identifier "GYOUMU_ID" paired with the table identifier "MST_GYOUMU".

In the conversion parameter acquisition means 347, the attribute value of the translation item name acquired by the translation item name acquisition means 344 is the entity acquired by the entity acquisition unit 33 from the table identified by the table identifier acquired by the table identifier acquisition means 345. The converted parameter which is the attribute value included in the matching record and is the attribute value of the primary key identifier acquired by the primary key identifier acquisition means 346 is acquired.

Specifically, the conversion parameter acquisition means 347 has the attribute value of the translation item name “SHAIN_NAME” acquired from the employee master identified by the acquired table identifier “MST_SHAIN” as the acquired entity “Otsuka”. Acquires the converted parameter "shain_code = 2" which is an attribute value included in the matching record and is an attribute value of the acquired primary key identifier "SHAIN_CODE".

In addition, the conversion parameter acquisition means 347 is an entity in which the attribute values of the two translation item names "TIME_CARD_DATE" and "SHAIN_NAME" acquired from the working hours table identified by the acquired table identifier "TRN_TIME_CARD" are acquired. It is an attribute value included in the record that matches the entity "April 1st to April 30th" that converted "last month" and the acquired entity "Otsuka", and the two acquired primary key identifiers "SHAIN_CODE". And the converted parameter "shan_code = 2, sta_date = 2019041, end_date = 20190430" which is the attribute value of "TIME_CARD_DATE" is acquired.

The API information acquisition unit 35 acquires the API information corresponding to the intent determined by the intent determination unit 31 from the API information storage unit 13.

The API information acquisition unit 35 acquires, for example, API information having an action name corresponding to the intent determined by the intent determination unit 31 from the API information storage unit 13.

Specifically, in the API information storage unit 13, for example, the action name "overtime hour inquiry" and three or more parameter specific information "employee code, share_code", "start date, sta_date", "end date, end_date", etc. API information 1 having the action name "working time data inquiry" and API information 2 having three or more parameter specific information "employee code, share_code", "work date, time_card_date", "business ID, gyoumu_id", etc. When the intent that is stored and is specified by the intent name "overtime inquiry intent" is acquired, the API information acquisition unit 35 has the API information 1 having the action name "overtime hour inquiry" that the intent has. To get.

The inquiry information configuration unit 36 configures inquiry information by using one or more parameters acquired by the parameter acquisition unit 34 and the API information acquired by the API information acquisition unit 35. Inquiry information is information for searching information, and is usually feasible information. The query information is, for example, a function or method in which an argument is inserted, but it may be a completed SQL statement or a set of URL and parameter.

The inquiry information configuration unit 36 is, for example, a parameter corresponding to each location at each of one or more variables of the API information acquired by the API information acquisition unit 35, and arranges the parameters acquired by the parameter acquisition unit 34. By doing so, inquiry information is constructed.

Specifically, for example, the inquiry information "https: /" in which "shain_code = 2", sta_date = 2019041 ", and end_date" = 20190430 are arranged at each of the three or more variables of the acquired API information 1. /Develop.soppra ...? Shan_code = 2 & sta_date = 2019041 & end_date = 20190430 "may be configured.

The search unit 37 executes the inquiry information configured by the inquiry information configuration unit 36 and acquires the search result. For example, the inquiry information "https://develop.soppra ...? Shanin_code = 2 & sta_date = 2019041 & end_date = 20190430" is executed, and the employee "Jiro Otsuka" specified by employee code 2 is executed from April 1 to April 30 Overtime hours "0401 = 1,0402 = 0, ... 4030 = 2" corresponding to each date up to the day may be acquired. Inquiry information such as SQL and detailed operations of the search unit 37 will be described with specific examples and modified examples.

The output unit 4 outputs various information. The various types of information are, for example, search results.

For example, the output unit 4 obtains information such as a search result, which is the result of various processing performed by the processing unit 3 in response to the reception unit 2 receiving information such as a conversation sentence in pairs with the terminal identifier. Send to the terminal identified by the identifier. Alternatively, for example, in response to the reception unit 2 receiving information such as a conversational sentence via an input device such as a touch panel or a microphone, the output unit 4 outputs information such as a search result to an output device such as a display or a speaker. It may be output via.

However, the output unit 4 may print out various types of information with a printer, store it in a recording medium, pass it on to another program, or transmit it to an external device, and the output unit 4 may output the output. The mode does not matter.

The search result output unit 41 outputs the search result acquired by the search unit 37. The search result output unit 41 transmits, for example, the search result acquired by the search unit 37 in response to the conversation sentence reception unit 21 receiving the conversation sentence in pairs with the terminal identifier to the terminal identified by the terminal identifier. .. Alternatively, the search result output unit 41 outputs the search result acquired by the search unit 37 in response to the conversation sentence reception unit 21 receiving the conversation sentence via an input device such as a microphone on a display, a speaker, or the like. It may be output via a device.

Storage unit 1, table storage unit 11, intent storage unit 12, API information storage unit 13, conversational text information storage unit 14, entity storage unit 15, entity mapping information storage unit 16, PK item storage unit 17, and day conversion information. The storage unit 18 is preferably a non-volatile recording medium such as a hard disk or a flash memory, but can also be realized by a volatile recording medium such as a RAM.

The process of storing information in the storage unit 1 etc. does not matter. For example, information may be stored in the storage unit 1 or the like via a recording medium, or information transmitted via a network, a communication line, or the like may be stored in the storage unit 1 or the like. Well, or the information input via the input device may be stored in the storage unit 1 or the like. The input device may be, for example, a keyboard, a mouse, a touch panel, a microphone, or the like. The above items regarding the storage unit 1 and the like also apply to the storage unit 10 and the like described in the second embodiment.

The storage unit 1 and the like may be outside the information output device 500. That is, the information such as the table may be stored in an external recording medium accessible to the computer of the information output device 500. Further, such a matter also applies to the storage unit 10 and the like described in the second embodiment.

The reception unit 2, the conversation text reception unit 21, the voice reception means 211, and the voice recognition means 212 may or may not include the input device. The reception unit 2 and the like can be realized by the driver software of the input device, or by the input device and its driver software. It should be noted that such a matter also applies to the reception unit 20 described in the second embodiment.

Processing unit 3, intent determination unit 31, conversational text information determination unit 32, entity acquisition unit 33, parameter acquisition unit 34, API information acquisition unit 35, inquiry information configuration unit 36, search unit 37, judgment means 341, day information The acquisition means 342, the entity name acquisition means 343, the translation item name acquisition means 344, the table identifier acquisition means 345, the primary key identifier acquisition means 346, and the conversion parameter acquisition means 347 can usually be realized from an API, a memory, or the like. The processing procedure of the processing unit 3 and the like is usually realized by software, and the software is recorded on a recording medium such as ROM. However, the processing procedure may be realized by hardware (dedicated circuit). It should be noted that such a matter also applies to the processing unit 30 and the like described in the second embodiment and the intent generation unit 301a and the like described in the third embodiment.

The output unit 4 and the search result output unit 41 may or may not include output devices such as displays and speakers. The output unit 4 and the like can be realized by the driver software of the output device, or by the output device and its driver software. It should be noted that such a matter also applies to the output unit 40 described in the second embodiment.

The receiving function of the reception unit 2 or the like is usually realized by a wireless or wired communication means (for example, a communication module such as a NIC (Network interface controller) or a modem), but a means for receiving a broadcast (for example, a broadcast). It may be realized by the receiving module). It should be noted that such a matter also applies to the reception function of the reception unit 20 described in the second embodiment. Further, the transmission function of the output unit 4 or the like is usually realized by a wireless or wired communication means, but may be realized by a broadcasting means (for example, a broadcasting module). It should be noted that such a matter also applies to the transmission function of the output unit 40 described in the second embodiment.

Next, the operation of the information output device 500 will be described with reference to the flowcharts of FIGS. 2 and 3.

FIG. 2 is a flowchart illustrating the operation of the information output device 500. When a conversational sentence is transmitted from one or more terminals to a pair with the terminal identifier, the flowcharts of FIGS. 2 and 3 are executed for each one or more terminal identifiers.

(Step S201) The processing unit 3 determines whether or not the conversation sentence reception unit 21 has accepted the conversation sentence. If it is determined that the conversation sentence reception unit 21 has accepted the conversation sentence, the process proceeds to step S202, and if it is determined that the conversation sentence has not been accepted, the process returns to step S201.

(Step S202) The intent determination unit 31 determines an intent corresponding to the conversational sentence received in step S201 from one or more intents stored in the intent storage unit 12.

(Step S203) The conversation sentence information determination unit 32 determines the conversation sentence information most similar to the conversation sentence received in step S201 from one or more conversation sentence information corresponding to the intent determined in step S202. To do.

(Step S204) The entity acquisition unit 33 corresponds to each one or more entities corresponding to the conversational sentence information determined in step S203, and stores one or more entities of the conversational sentence received in step S201. Obtained from part 15.

(Step S205) The parameter acquisition unit 34 executes a process of acquiring one or more parameters corresponding to one or more entities acquired in step S204 (hereinafter, may be referred to as a parameter acquisition process). The parameter acquisition process will be described with reference to FIG.

(Step S206) The API information acquisition unit 35 acquires the API information corresponding to the intent determined in step S202 from the API information storage unit 13.

(Step S207) The inquiry information configuration unit 36 configures inquiry information by using one or more parameters acquired in step S205 and the API information acquired in step S206.

(Step S208) The search unit 37 executes the inquiry information configured in step S207 and acquires the search result.

(Step S209) The search result output unit 41 outputs the search result acquired in step S208. After that, the process returns to step S201.

In the flowchart of FIG. 2, the process starts when the power of the information output device 500 is turned on or the program is started, and the process ends when the power is turned off or an interrupt for the end of the process occurs. However, the trigger for the start or end of processing does not matter.

FIG. 3 is a flowchart illustrating the parameter acquisition process in step S205.

(Step S301) The parameter acquisition unit 34 sets the initial value 1 in the variable i. The variable i is a variable for sequentially selecting unselected entities from one or more entities acquired in step S204.

(Step S302) The parameter acquisition unit 34 determines whether or not there is an i-th entity. If it is determined that there is an i-th entity, the process proceeds to step S303, and if it is determined that there is no i-th entity, the process returns to higher-level processing.

(Step S303) The entity name acquisition means 343 acquires the entity name corresponding to the i-th entity from the entity storage unit 15.

(Step S304) The translation item name acquisition means 344 acquires the translation item name paired with the entity name acquired in step S303 from the entity mapping information storage unit 16.

(Step S305) The table identifier acquisition means 345 acquires a table identifier paired with the entity name acquired in step S303 from the entity mapping information storage unit 16.

(Step S306) The primary key identifier acquisition means 346 acquires a primary key identifier paired with the table identifier acquired in step S305 from the PK item storage unit 17.

(Step S307) The determination means 341 determines whether or not the i-th entity is a day word. If the i-th entity is determined to be a day word, the process proceeds to step S308, and if it is determined that the entity is not a day word, the process proceeds to step S309.

(Step S308) The day information acquisition means 342 acquires the day conversion information corresponding to the day word from the day conversion information storage unit 18, and uses the acquired day conversion information to obtain the parameter day information (start date, end). Get the date etc.).

(Step S309) The conversion parameter acquisition means 347 is an attribute value included in the record in which the attribute value of the translation item name matches the i-th entity from the table identified by the table identifier acquired in step S305. The converted parameters (Shain_code = 1, sta_date = 2019041, end_date = 20190430, etc.), which are the attribute values of the primary key identifier acquired in step S306, are acquired.

(Step S310) The parameter acquisition unit 34 increments the variable i. After that, the process returns to step S302.

Hereinafter, a specific operation example of the information output device 500 according to the present embodiment will be described. The following description can be changed in various ways and does not limit the scope of the present invention.

The information output device 500 in this example is a stand-alone terminal. For example, three table information as shown in FIG. 4 is stored in the table storage unit 11 of the information output device 500. FIG. 4 is a data structure diagram of table information. The table information has a table identifier and a table. An ID (for example, "1", "2", etc.) is associated with each stored table information.

For example, the table information corresponding to the ID "1" (hereinafter, may be referred to as table information 1) has a table identifier "MST_SHAIN" and a first table (employee master). In the employee master, the first item name "SHAIN_CODE" and the corresponding two or more values "1", "2" ..., And the second item name "SHAIN_CODE" and the corresponding two or more values " "Ichiro Nakamura", "Jiro Otsuka", etc. are registered.

Further, the table information 2 corresponding to the ID "2" has a table identifier "MST_GYOUMU" and a second table (business master). In the business master, the first item name "GYOUMU_ID" and the corresponding two or more values "a", "b" ..., And the second item name "GYOUMU_NAME" and the corresponding two or more values " "A project", "B project", etc. are registered.

Further, the table information 3 has a table identifier "TRN_TIME_CARD" and a third table (working time table). In the working hours table, the first item name "TRN_TIME_CARD" and the corresponding two or more values "4/1", "4/1" ... "4/2", "4/2" ... "4/30", "4/30" ..., the second item name "SHAIN_CODE" and the corresponding two or more values "1", "2" ... "1", "2" ... -"1", "2", the third item name "ZIKANGAI_TIME" and the corresponding two or more values "2", "1" ... "1", "0" ... "3", " 2 ”, the fourth item name“ TOTAL_TIME ”and the corresponding two or more values“ 10 ”,“ 9 ”・・・“ 9 ”,“ 8 ”・・・“ 11 ”,“ 10 ”, and the fifth The item name "GYOUMU_ID" and the corresponding two or more values "a", "b" ... "a", "b" ... "b", "b", etc. are registered.

For example, two or more intents as shown in FIG. 5 are stored in the intent storage unit 12. FIG. 5 is a data structure diagram of the intent. The intent has an intent name and an action name. An ID (for example, "1", "2", etc.) is associated with each stored intent.

For example, the intent 1 corresponding to the ID "1" has an intent name "overtime inquiry intent" and an action name "overtime hour inquiry". In addition, the intent 2 has an intent name “work inquiry data inquiry intent” and an action name “work inquiry data inquiry”.

The API information storage unit 13 stores, for example, a set of pairs of action names and API information as shown in FIG. FIG. 6 is a diagram showing an example of the contents of the API information storage unit 13. The pair of action name and API information is, for example, the pair of action name "overtime hours inquiry" and API information "search_overtime hours (employee code, start date, end date)", and the action name "working hours data inquiry". And the API information "service_working hours (employee code, business date, business ID)".

For example, as shown in FIG. 7, the conversation sentence information storage unit 14 stores one or more conversation sentence information in association with each one or more intents, and further, each one or more conversation sentences. One or more entity information is also stored in association with the information. FIG. 7 is a diagram showing an example of the contents of the conversation text information storage unit 14. The entity information has an entity, a start position, an end position, and an entity name.

For example, the first intent "Overtime time inquiry intent" is associated with five conversational sentence information 1-5. Of these, conversational text information 1 is "tell me the overtime hours of Nakamura last month", conversational text information 2 is "tell me the overtime hours of Nakamura last month", and conversational text information 3 is "tell me the overtime hours of Nakamura last month". "Tell me the overtime hours of Nakamura", the conversational sentence information 4 is "tell me the overtime hours of last month", and the conversational sentence information 5 is "tell me the overtime hours".

The conversation text information 1 corresponding to the "overtime hour inquiry intent" includes the first entity information having the entity "Nakamura", the start position "1", the end position "2", and the entity name "employee entity". , Entity "last month", start position "4", end position "5", and second entity information with entity name "date entity", entity "overtime", start position "7", end position "10" , And a third entity information with the entity name “overtime entity”.

Further, the conversation text information 2 includes a fourth entity information having an entity "last month", a start position "1", an end position "2", and an entity name "date entity", and an entity "Nakamura", a start position ". Fifth entity information with 4 ”, end position“ 5 ”, and entity name“ employee entity ”, entity“ overtime hours ”, start position“ 7 ”, end position“ 10 ”, and entity name“ overtime entity ” Corresponds to the sixth entity information having.

Further, the conversation text information 3 includes a seventh entity information having an entity "Nakamura", a start position "1", an end position "2", and an entity name "employee entity", and an entity "overtime hours" and a start position. The eighth entity information having "4", the end position "7", and the entity name "overtime entity" is associated with each other.

Further, the conversation text information 4 includes a ninth entity information having an entity "last month", a start position "1", an end position "2", and an entity name "date entity", and an entity "overtime hours" and a start position. The tenth entity information having "4", the end position "7", and the entity name "overtime entity" is associated with each other.

Further, the conversation text information 5 is associated with eleventh entity information having an entity "overtime hours", a start position "1", an end position "4", and an entity name "overtime entity".

On the other hand, the second intent "working time data inquiry intent" is associated with, for example, 15 conversational sentence information 1 to 15. Of these, conversation sentence 1 is "Tell me the working hours of Ichiro Nakamura's A project yesterday", and conversation sentence 2 is "Tell me the working hours of Ichiro Nakamura's A project yesterday", and the last Conversational sentence 15 is "Tell me the working hours". In addition, the conversation sentence 2 and the following are sentences in which the word order is changed or some words are omitted from the conversation sentence 1.

The conversation text information 1 corresponding to the "working time data inquiry intent" is the first entity having the entity "Ichiro Nakamura", the start position "1", the end position "4", and the entity name "employee entity". Second entity information with information and entity "yesterday", start position "6", end position "7", and entity name "date entity", entity "working hours", start position "13", end position Corresponds to "16" and a third entity information having the entity name "working time entity". It should be noted that each of the conversation sentences 2 to 15 is associated with one or more entity information similarly configured.

In the entity storage unit 15, for example, as shown in FIG. 8, one or two or more entities are stored in association with each one or more entity names. FIG. 8 is a diagram showing an example of one or more entities (hereinafter, may be referred to as an entity group) corresponding to each entity name. As described above, the pair of the entity name and the entity group may be considered as the entity information.

For example, the entity group corresponding to the entity name "employee entity" includes "Ichiro Nakamura", "Jiro Otsuka" and the like. The entity group corresponding to the entity name "date entity" includes "last month", "yesterday" and the like. The entity group corresponding to the entity name "overtime entity" includes "overtime hours", "overtime" and the like. The entity group corresponding to the entity name "business entity" includes "A matter", "B matter" and the like. The entity group corresponding to the entity name "working time entity" includes "working time", "working" and the like.

The entity mapping information storage unit 16 stores, for example, two or more entity mapping information (entity mapping information 1 to 4, etc.) as shown in FIG. FIG. 9 is a data structure diagram of entity mapping information. The entity mapping information has a table identifier, a translation item name, and an entity name.

Note that entity mapping information 1 to 3 will not be repeated as described above. The entity mapping information 4 has a table identifier "MST_GYOUMU", a translation item name "GYOUMU_ID", and an entity name "business entity".

For example, two or more PK items (PK items 1 to 4, etc.) are stored in the PK item storage unit 17, as shown in FIG. FIG. 10 is a data structure diagram of PK items. The PK item has a table identifier and a primary key identifier. Since PK items 1 to 4 have been described above, they will not be repeated.

The day conversion information storage unit 18 stores a set (not shown) of pairs of day words and day information acquisition information. The pair of the day word and the day information acquisition information is, for example, a pair of the day word "last month" and the day information acquisition information 1, a pair of the day word "yesterday" and the day information acquisition information 2, and the like. For example, the day information acquisition information 1 corresponding to the day word "last month" acquires the current time information, acquires the previous month with respect to the month possessed by the current time information, and refers to the calendar information of the previous month. Then, it is a program to acquire the day information from the first day to the last day of the previous month. The day information acquisition information 2 corresponding to the day word "yesterday" is a method for acquiring the current time information and acquiring the day information of the day before the day held by the current time information.

The current time is 11:15 on May 10, 2019. Now, suppose that the user voice-inputs "Tell me about Otsuka's overtime hours last month" to the terminal.

In the information output device 500, the voice receiving means 211 receives the above voice via the microphone, the voice recognition means 212 performs voice recognition processing on the voice, and the conversation sentence "Tell me the overtime hours of Otsuka last month". To get.

The intent determination unit 31 determines the intent corresponding to the conversation sentence from the two or

more intents

1, 2, ... Stored in the intent storage unit 12. Specifically, the intent determination unit 31 acquires three independent words "Otsuka", "last month" and "overtime hours" from the conversation sentence by, for example, morphological analysis, and uses each independent word as a key to store the intent. 12 is searched to determine an intent 1 having an action name “overtime hours reference” that partially matches “overtime hours”.

The conversational sentence information determination unit 32 refers to the conversational sentence information storage unit 14 (FIG. 7), and is the closest to the conversational sentence from the five conversational sentence information 1 to 5 corresponding to the determined intent 1. Determine the conversational text information to be used. Here, the conversation sentence information 1 is determined, in which only the employee entity "Nakamura" specified by the start position "1" and the end position "2" is different from the conversation sentence, and the other entities are the same. Employee.

The entity acquisition unit 33 is an entity corresponding to the three entities "Nakamura", "last month", and "overtime hours" corresponding to the determined conversation sentence information 1, and is the three entities possessed by the conversation sentence. A certain "Otsuka", "last month", and "overtime hours" are acquired from the entity storage unit 15.

Specifically, the conversation sentence information determination unit 32 refers to the first entity information among the first to third entity information corresponding to the conversation sentence information 1, and starts the start position “1” in the conversation sentence. And it is determined that "Otsuka" specified by the end position "2" is an employee entity. Next, the conversation sentence information determination unit 32 refers to the entity storage unit 15 and acquires the entity "Otsuka Jiro" that partially matches "Otsuka" from the entity group corresponding to the entity name "employee entity".

Further, the conversation sentence information determination unit 32 refers to the second entity information and determines that the "last month" specified by the start position "4" and the end position "5" in the conversation sentence is the date entity. , Get the entity "last month" that exactly matches "last month" from the entity group corresponding to the entity name "date entity".

Further, the conversation sentence information determination unit 32 determines that the “overtime hours” specified by the start position “7” and the end position “10” in the conversation sentence is the overtime entity by referring to the third entity information. Then, from the entity group corresponding to the entity name "overtime entity", the entity "overtime hours" that exactly matches the "overtime hours" is acquired.

In this way, when the three entities "Jiro Otsuka", "last month", and "overtime hours" corresponding to the conversation sentence are acquired, the parameter acquisition unit 34 acquires the three parameters corresponding to the three entities. ..

Specifically, first, regarding the first entity "Jiro Otsuka", the entity name acquisition means 343 constituting the parameter acquisition unit 34 uses the first entity information to transfer the entity name "employee entity" from the entity storage unit 15. get. Next, the translation item name acquisition means 344 acquires the translation item name “SHAIN_NAME” paired with the acquired entity name “employee entity” from the entity mapping information storage unit 16 using the entity mapping information 1. Further, the table identifier acquisition means 345 acquires the table identifier "MST_SHAIN" paired with the acquired entity name "employee entity" from the entity mapping information storage unit 16 using the entity mapping information 1.

Next, the primary key identifier acquisition means 346 acquires the primary key identifier "SHAIN_CODE" paired with the acquired table identifier "MST_SHAIN" from the PK item storage unit 17 using the PK item 1. Further, the determination means 341 determines that the first entity "Otsuka" is not a Japanese word, and the conversion parameter acquisition means 347 corresponds to the table (employee master) identified by the acquired table identifier "MST_SHAIN". Therefore, the attribute value of the translation item name "SHAIN_NAME" is the attribute value included in the record matching the first entity "Jiro Otsuka", and is the attribute value of the acquired primary key identifier "SHAIN_CODE". Acquires the converted parameter "shain_code = 2" of the eye.

Next, regarding the second entity "last month", the entity name acquisition means 343 acquires the entity name "date entity" corresponding to the "last month", and the translation item name acquisition means 344 acquires the acquired entity name "date". The translation item name "TIME_CARD_DATE" paired with "entity" is acquired by using the entity mapping information 2. The table identifier acquisition means 345 acquires the table identifier “TRN_TIME_CARD” paired with the acquired entity name “date entity” from the entity mapping information storage unit 16 using the entity mapping information 2.

Next, the primary key identifier acquisition means 346 acquires two primary key identifiers "SHAIN_CODE" and "TIME_CARD_DATE" that are paired with the acquired table identifier "TRN_TIME_CARD" using the two

PK items

2 and 3. The determination means 341 determines that the second entity "last month" is a day word, and in response to this, the day information acquisition means 342 stores the day conversion information corresponding to the day word "last month". The day information "4/1 to 4/30" is acquired by using the acquired day conversion information acquired from the unit 18.

Specifically, the day information acquisition means 342 acquires the current time information "May 10, 2019 11:15" from the built-in clock or the like, and April, which is the month before "May" in the current time information. With reference to the calendar information of, the day information "4/1 to 4/30" indicating 30 days from the start date "4/1" to the end date "4/30" is acquired.

The conversion parameter acquisition means 347 has a start date in which the attribute value of the translation item name "TIME_CARD_DATE" corresponds to the second entity "last month" from the table (working time table) identified by the acquired table identifier "TRN_TIME_CARD". The second and the attribute values included in the records that match "4/1" and the end date "4/30", which are the attribute values of the two acquired primary key identifiers "SHAIN_CODE" and "TIME_CARD_DATE". Acquire the third converted parameters "sta_date = 2019041" and "end_date = 20190430".

Next, regarding the third entity "overtime hours", the entity name acquisition means 343 acquires the entity name "overtime entity" corresponding to the "overtime hours", and the translation item name acquisition means 344 obtains the acquired entity name. The translation item name "ZIKANGAI_TIME" paired with the "overtime entity" is acquired using the entity mapping information 3. Further, the table identifier acquisition means 345 acquires the table identifier "TRN_TIME_CARD" paired with the acquired entity name "overtime entity" by using the entity mapping information 3.

Since the two primary key identifiers "SHAIN_CODE" and "TIME_CARD_DATE" that are paired with the acquired table identifier "TRN_TIME_CARD" have already been acquired, the primary key identifier acquisition means 346 normally does not acquire them. You may get it again.

Further, the determination means 341 determines that the third entity "overtime hours" is not a day word, and the conversion parameter acquisition means 347 has a translation item name of the working hours table identified by the acquired table identifier "TRN_TIME_CARD". Acquire the fourth converted parameter "zikangai_time" corresponding to "ZIKANGAI_TIME".

In this way, when the four parameters "shain_code = 2", "sta_date = 2019041", "end_date = 20190430" and "zikangai_time" corresponding to the conversation sentence "Tell me about the overtime hours of Otsuka last month" are acquired, the API The information acquisition unit 35 acquires the API information “sent_overtime hours (employee code, start date, end date)” corresponding to the action name “overtime time inquiry” possessed by the determined intent 1 from the API information storage unit 13. ..

The inquiry information configuration unit 36 substitutes the acquired four parameters into the corresponding variables of the acquired API information "search_overtime hours (employee code, start date, end date)" to obtain inquiry information. It constitutes "service_zikangai_time (shain_code = 2, sta_date = 2019041, end_date = 20190430".

The search unit 37 executes the configured inquiry information and acquires the search result. Here, from the overtime hours table identified by the table identifier "TRN_TIME_CARD", the value of the item name "TIME_CARD_DATE" is "4/1" to "4/30", and the value of SHAIN_CODE is "2" 30. From the records, 30 values "1,0, ... 2" corresponding to the item name "ZIKANGAI_TIME" are acquired, and the search result "20 hours" obtained by totaling the 30 values is acquired.

The search result output unit 41 outputs a response sentence including the acquired search result "20 hours". The output response text may be, for example, "20 hours" or "Otsuka-sama's overtime hours in April is 20 hours". Specifically, for example, the template of the response sentence is stored in the storage unit 1, and the search result output unit 41 sets the variable of the template to the above search result "20 hours" and "obtained from the conversation sentence". The response sentence may be constructed by substituting "Otsuka" and "April" obtained in the search process.

Although detailed explanation is omitted, for the conversational sentence "Tell me yesterday's working hours of A project of Jiro Otsuka", for example, four parameters "shain_code = 2", "time_card_date = 20190509", " “Gyoumu_id = a” and “total_time” may be acquired, and the inquiry information “sent_total_time (sent_code = 2, time_card_date = 20109009, gyoumu_id = a” may be configured.

(Modification 1) When the API information is the SQL statement "select_overtime_from_table name_where_employee code, business date", the parameter acquisition unit 34 has four parameters "shain_code = 2" and "time_caed_date = 20190509". , "Trn_time_card", and "zikangai_time" are acquired, and the inquiry information configuration unit 36 may configure the inquiry information "select_zikangai_time_from_trans_time_card_where_shine_code = 2, time_code = 2, time_code = 2, time_code".

(Modification 2) When the API information is a set of URL and method "http://develop.soppra ...? Employee code & start date & end date", the parameter acquisition unit 34 has four parameters " "Shin_code = 2", "sta_date = 2019041", "end_date = 20190430", and "zikangai_time" are acquired, and the inquiry information configuration unit 36 acquires the inquiry information "http://develop.soppra ...? Shan_code =" 2 & sta_date = 2019041 & end_date = 20190430 ”may be configured.

As described above, according to the present embodiment, one or more intents, which are information managed for each business process and have an action name for specifying the business process, are stored in the intent storage unit 12, and the API is The information storage unit 13 stores one or more API information, which is information corresponding to the intent, information for performing an information search, and information having information for specifying a parameter which is one or more variables. The conversation sentence information storage unit 14 stores one or more conversation sentence information which is information corresponding to the intent and is the conversation sentence information, and the entity storage unit 15 stores any one or more conversation sentence information. One or more entities that are words to be associated with each other are stored, and the information output device 500 receives a conversation sentence, determines an intent corresponding to the received conversation sentence, and one or more conversation sentence information corresponding to the determined intent. From among them, determine the conversational sentence information that most closely resembles the received conversational sentence, and correspond to one or more entities corresponding to the determined conversational sentence information, and one or more words that the received conversational sentence has. Acquire the entity, acquire one or more parameters corresponding to each of the acquired one or more entities, acquire the API information corresponding to the determined intent from the API information storage unit 13, and acquire the acquired one or more parameters. The same search result can be obtained by constructing the inquiry information for information search using the acquired API information, executing the configured inquiry information, acquiring the search result, and outputting the acquired search result. Information can be searched by accepting dynamically changing conversational sentences even for the requested inquiry.

Further, the information output device 500 acquires a text corresponding to the received conversational sentence, acquires one or more independent words from the text, and has an intent having an action name having a word that is the same as or similar to one or more independent words. By deciding, it is possible to accept dynamically changing conversational sentences and search for information even with a query for the same search result.

Further, in the day conversion information storage unit 18, the day word which is a word related to the day and the day information which is the information about the day corresponding to the day word and which is the information used when constructing the inquiry information is acquired. One or more day conversion information having the day information acquisition information is stored, and the information output device 500 acquires the day conversion information corresponding to the day word in the acquired one or more entities from the day conversion information storage unit 18. By acquiring the day information, which is a parameter, using the day conversion information, it is possible to accept dynamically changing conversational sentences even in a query for the same search result, and to appropriately search for information.

In the above configuration, the information output device 500 determines whether or not the day word exists in one or more acquired entities, and when it is determined that the day word exists, the day conversion corresponding to the day word is performed. The information may be acquired from the day conversion information storage unit 18, and the day information which is a parameter may be acquired by using the day conversion information. As a result, it is possible to accept dynamically changing conversational sentences even in a query for the same search result regardless of the presence or absence of a Japanese word, and to appropriately search for information.

Further, the entity of the entity storage unit 15 corresponds to the entity name which is the name of the entity, and the entity mapping information storage unit 16 has a table identifier for specifying the table and an attribute name in the table identified by the table identifier. A table in which one or more entity mapping information having a translation item name and an entity name, which is information for identifying the entity to be converted, is stored, and is identified by a table identifier and a table identifier in the PK item storage unit 17. One or more PK items having a primary key identifier that identifies the primary key of the above are stored, and the information output device 500 acquires the entity name corresponding to the acquired entity, and the translation item name paired with the acquired entity name. Is acquired from the entity mapping information storage unit 16, the table identifier paired with the acquired entity name is acquired from the entity mapping information storage unit 16, and the primary key identifier paired with the acquired table identifier is obtained from the PK item storage unit 17. Conversion that the attribute value of the acquired translation item name from the acquired table identified by the acquired table identifier is the attribute value included in the record that matches the acquired entity and is the attribute value of the acquired primary key identifier. By acquiring the obtained parameters, it is possible to accept dynamically changing conversational sentences even in a query for the same search result, and to appropriately search for information.

Further, the information output device 500 is a parameter corresponding to each place in each place of one or more variables possessed by the acquired API information, and the same search is performed by arranging the acquired parameters and configuring the inquiry information. Information can be searched by accepting dynamically changing conversational sentences even when inquiring for results.

Further, the information output device 500 receives the voice of the conversational sentence, performs voice recognition processing on the voice, and acquires the conversational sentence which is a character string, so that the voice dynamically changes even in the inquiry for the same search result. You can search for information by accepting conversational sentences.

Further, the processing in the present embodiment may be realized by software. Then, this software may be distributed by software download or the like. Further, this software may be recorded on a recording medium such as a CD-ROM and disseminated. It should be noted that this also applies to other embodiments herein.

The software that realizes the information output device 500 according to the present embodiment is, for example, the following program. That is, this program corresponds to the intent storage unit 12 and the intent that store one or more intents that are information managed for each business process and have an action name that specifies the business process. Corresponds to the API information storage unit 13 and the intent in which one or more API information is stored, which is information that exists, is information for performing an information search, and is information that has information that specifies a parameter that is one or more variables. Conversational sentence information storage unit 14 that stores one or more conversational sentence information that is attached information and is conversational sentence information, and one or more entities that are words corresponding to any one or more conversational sentence information are stored. The computer that can access the entity storage unit 15 is the conversation sentence reception unit 21 that accepts conversation sentences, the intent determination unit 31 that determines the intent corresponding to the conversation sentence received by the conversation sentence reception unit 21, and the above. Conversational sentence information determination unit 32 that determines the conversational sentence information that most closely resembles the conversational sentence received by the conversational sentence reception unit 21 from among one or more conversational sentence information corresponding to the intent determined by the intent determination unit 31. And one or more entities corresponding to each one or more entities corresponding to the conversational sentence information determined by the conversational sentence information determination unit 32, and one or more entities which are words possessed by the conversational sentence received by the conversational sentence reception unit 21. Corresponds to the entity acquisition unit 33 to be acquired, the parameter acquisition unit 34 that acquires one or more parameters corresponding to one or more entities acquired by the entity acquisition unit 33, and the intent determined by the intent determination unit 31. The API information acquisition unit 35 that acquires the API information to be generated from the API information storage unit 13, one or more parameters acquired by the parameter acquisition unit 34, and the API information acquired by the API information acquisition unit 35 are used. The inquiry information configuration unit 36 that constitutes inquiry information for information search, the search unit 37 that executes inquiry information configured by the inquiry information configuration unit 36 and acquires search results, and the search unit 37. This is a program for functioning as a search result output unit 41 that outputs the search results acquired by.

FIG. 11 is an external view of a computer system 900 that executes the program according to the present embodiment to realize the information output device 500. This embodiment can be realized by computer hardware and a computer program executed on the computer hardware. In FIG. 11, the computer system 900 includes a computer 901 including a disk drive 905, a keyboard 902, a mouse 903, a display 904, a microphone 917, and a speaker 918. The entire system including the keyboard 902, the mouse 903, the display 904, the microphone 917, and the speaker 918 may be called a computer.

FIG. 12 is a diagram showing an example of the internal configuration of the computer system 900. In FIG. 12, the computer 901 is connected to the MPU 911, the ROM 912 for storing a program such as a bootup program, and the MPU 911 in addition to the disk drive 905, and temporarily stores the instructions of the application program and temporarily. It provides a RAM 913 that provides a storage space, a storage 914 that stores application programs, system programs, and data, a bus 915 that interconnects the MPU 911, ROM 912, and the like, and a connection to a network such as an external network or an internal network. It includes a network card 916. The storage 914 is, for example, a hard disk, an SSD, a flash memory, or the like.

A program for causing the computer system 900 to execute the function of the information output device 500 may be stored in a disk 921 such as a DVD or a CD-ROM, inserted into the disk drive 905, and transferred to the storage 914. Alternatively, the program may be transmitted to computer 901 over the network and stored in storage 914. The program is loaded into RAM 913 at run time. The program may be loaded directly from disk 921 or the network. Further, the program may be read into the computer system 900 via another removable recording medium (for example, a DVD, a memory card, etc.) instead of the disk 921.

The program does not necessarily have to include an operating system (OS) for executing the function of the information output device 500, a third-party program, or the like in 901 showing the details of the computer. The program may contain only a portion of instructions that call the appropriate function or module in a controlled manner to achieve the desired result. It is well known how the computer system 900 works, and detailed description thereof will be omitted.

The computer system 900 described above is a server or a stationary PC, but the information output device 500 may be realized by a mobile terminal such as a tablet terminal, a smartphone, or a notebook PC. In this case, for example, the keyboard 902 and the mouse 903 may be replaced with a touch panel, the disk drive 905 may be replaced with a memory card slot, and the disk 921 may be replaced with a memory card. However, the above is an example, and the hardware configuration of the computer that realizes the information output device 500 does not matter.

In the above program, in the transmission step of transmitting information and the receiving step of receiving information, processing performed by hardware, for example, processing performed by a modem or interface card in the transmission step (only performed by hardware). Processing that is not done) is not included.

Further, the number of computers that execute the above program may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed.

The computer system 900 described above can also realize the information generation devices of the second and third embodiments described below and the information processing devices of the fourth and fifth embodiments.

(Embodiment 2)
The information generation device 100 in the present embodiment is communicably connected to the information output device 500 described in the first embodiment via, for example, a network or a communication line. Like the information output device 500, the information generation device 100 is also a server in a company that provides services such as information retrieval, but it may be a stand-alone terminal, and the means for realizing the information may be limited. It should be noted that such matters are also common to the third embodiment.

The information generation device 100 is a device that generates information such as an intent, one or more conversational sentence information, and one or more entities used by the information output device 500 described in the first embodiment. The generated information such as one or more conversational text information and one or more entities is passed or transmitted from the information generation device 100 to the information output device 500, and the intent storage unit 12 of the information output device 500, the conversation. It is stored in the statement information storage unit 14, the entity storage unit 15, and the like.

FIG. 13 is a block diagram of the information generation device 100 according to the present embodiment. The information generation device 100 includes a storage unit 10, a reception unit 20, a processing unit 30, and an output unit 40. The storage unit 10 stores the document storage unit 101, the template storage unit 102, the intent storage unit 12a, the API information storage unit 13a, the conversation text information storage unit 14a, the entity storage unit 15a, the entity mapping information storage unit 16a, and the PK item storage. A portion 17a is provided. The table storage unit 11a may be considered to be included in the document storage unit 101. That is, the table may be considered as a kind of document described later.

The processing unit 30 includes an intent generation unit 301, an entity generation unit 302, a conversational sentence information generation unit 303, an entity mapping information generation unit 304, a PK item generation unit 305, and a storage unit 306. The intent generation unit 301 includes API information extraction means 3011, action name acquisition means 3012, and intent generation means 3013. The entity generation unit 302 includes an entity name generation means 3021 and an entity generation means 3022. The conversation sentence information generation unit 303 includes an entity acquisition unit 3031 and a conversation sentence generation unit 3032. The storage unit 306 includes an intent storage means 3061, an entity storage means 3062, a conversational sentence information storage means 3063, an entity mapping information storage means 3064, and a PK item storage means 3065.

Various information is stored in the storage unit 10. The various types of information include, for example, documents, design documents, templates, tables, intents, conversational text information, entities, entity mapping information, PK items, and the like. The document, API design document, and template will be described later. Further, the table, the intent, the conversational sentence information, the entity, the entity mapping information, and the PK item have been described in the first embodiment and will not be repeated.

One or two or more documents are stored in the document storage unit 101 that constitutes the storage unit 10. A document is a set of information processed by a computer, and may be said to be a set of information including one or more character strings constituting a sentence or word in a natural language.

It is preferable that the document contains, for example, an action name or a character string having a part of the action name. Since the action name has been described in the first embodiment, it will not be repeated. In the following, a document containing an action name or a character string having a part of the action name may be referred to as a "first document".

The first document is, for example, a document file. A document file is a file related to a document. The document is, for example, a design document. The design document is a document related to software design, and may be referred to as a specification document, a definition document, or the like. The software is, for example, API, but may be a database or the like, and the type thereof does not matter. The API and database have been described above and will not be repeated. The design document is, for example, an API design document described later, but may be a DB design document. The DB design document will be described in the third embodiment. However, the document is not limited to the design document, and may be an in-house document such as a daily report, a business instruction, or a plan, or a document exchanged with the outside such as an order form or a delivery note. It doesn't matter.

As mentioned above, the document file is, for example, a file such as Word or Excel, but it may be hypertext, and its format does not matter. Hypertext is a document written in a markup language such as HTML or XML, and can be said to be a document file having a set (also referred to as a link) for associating words in the document with other documents. ..

Alternatively, the document may have, for example, an attribute name and one or more words corresponding to the attribute name. The attribute names, etc. will not be repeated as described above. The attribute name and one or more words corresponding to the attribute name may be said to be, for example, an entity name and a word capable of generating one or more entities corresponding thereto. In the following, a document having an attribute name and one or more words corresponding to the attribute name may be referred to as a "second document". However, the second document may have, for example, only the attribute name and may not have one or more words corresponding to the attribute name.

The second document is, for example, a database other than one or more tables stored in the table storage unit 11a. However, it may be considered that the second document also includes one or more tables stored in the table storage unit 11a. Alternatively, the second document may be, for example, a document file, regardless of its type. For example, the first document may be hypertext and the second document may be one or more document files linked to one or more words in the first document.

However, the first document and the second document may be one document. The first document and the second document may be, for example, one document file. That is, one document (for example, one document file) may have an action name or a part of the action name and an attribute name, and further, one or more words corresponding to the attribute name. You may also have. A specific example 2 is a document file which is an example of a document having an action name or a part of the action name and an attribute name, and an operation example of generating a corpus such as an intent, an entity name, and a template from the document file. It will be explained in.

Further, it is preferable that either the first or second document has a tabular format, but it is not necessary to have a tabular format. A document having a tabular format is, for example, a document created by spreadsheet software such as Excel, but a document created by a document creation software such as Word may be converted into a tabular format, and the format and creation method may be used. Does not matter.

The document that does not have a tabular format is, for example, a Word document file, but it may be a document file such as an email or chat. The document file such as a word is preferably structured, for example, but does not have to be structured. What is structured is that it is described in a markup language such as HTML or XML. Alternatively, the structuring may be realized by, for example, adding a heading or a number, inserting a symbol or a space, a line break, or the like, and the method of realizing the structure does not matter.

In the present embodiment, the document stored in the document storage unit 101 is, for example, an API design document. The API design document is a design document related to API. The API design document has, for example, one or more API information. Since the API information has been described in the first embodiment, it will not be repeated. The API information usually corresponds to the action name.

The API design document may include, for example, a set of pairs of action names and API information as shown in FIG. Note that FIG. 6 and the pair of the action name and the API information will not be repeated because they have been described in the first embodiment.

In addition, the API design document usually includes information such as parameter names, table identifiers, and primary key identifiers. The parameter name, table identifier, primary key identifier, etc. are not repeated as described above. However, the structure of the API design document does not matter.

One or two or more templates are stored in the template storage unit 102. The stored template is a conversational sentence template and has one or more variables. For example, as described in the first embodiment, if the variable is "{}", the template may be, for example, "Tell me {} of {} of {}".

In the template storage unit 102, for example, one or more templates may be stored in association with the action name. However, the information associated with one or more templates may be, for example, an intent name, a table name, an attribute name, or the like.

It is preferable that each one or more variables of the template are associated with entity names such as "employee entity", "date entity", and "overtime entity". Specifically, for example, if the variables associated with the entity names are "{employee entity}", "{date entity}", and "{overtime entity}", the template is, for example, {{employee entity} { Tell me the {overtime entity} of the date entity}.

The template storage unit 102 is, for example, a template having one or more variables associated with entity names as shown in FIG. 14, and two or more templates having different positions or numbers of variables are, for example, action names. Can be stored in association with.

Two or more templates corresponding to the action name "Overtime time inquiry" are, for example, the first template "Tell me the {Overtime entity} of the {Date entity} of the {Employee entity}" and the second template "{Date". Tell me {overtime entity} of {employee entity} of entity}, "tell me {overtime entity} of {employee entity}", {overtime entity} of the fourth template "{date entity}" Tell me ", and the fifth template" Tell me {overtime entity} ".

Two or more templates corresponding to the action name "working time data inquiry" are, for example, the first template "Tell me the {working time entity} of the {business entity} of the {date entity} of the {employee entity}", The second template "Tell me the {working time entity} of the {date entity} of the {business entity} of the {employee entity}", and the {business entity} of the {employee entity} of the third template "{date entity}" "Tell me {working time entity}", 4th template "Tell me {working time entity} of {employee entity} of {business entity} of {date entity}", 5th template "{business entity} Tell me the {working time entity} of the {employee entity} of the {date entity} ", the sixth template" tell me the {working time entity} of the {date entity} of the {employee entity} of the {business entity} ", Seventh template "Tell me {working time entity} of {date entity} of {employee entity}", eighth template "tell {working time entity} of {business entity} of {employee entity}", Ninth template "Tell me {working time entity} of {employee entity} of {date entity}", tenth template "tell {working time entity} of {business entity} of {date entity}", Tell me the {working time entity} of the {date entity} of the eleventh template "{business entity}" and the {working time entity} of the {employee entity} of the twelfth template "{business entity}" "Tell me the {working time entity} of the thirteenth template" {employee entity} ", the fourteenth template" tell me the {working time entity} of the {date entity} ", the fifteenth template" It may be "Tell me {working time entity} of {business entity}" and "Tell me {working time entity}" of the sixteenth template.

In the template storage unit 102, for example, a template (hereinafter, may be referred to as an initial template) in which an entity name does not correspond to each variable, such as "Tell me {} of {} of {}" at the beginning, is stored. The processing unit 30 is generated in response to the fact that the entity generation unit 302 described later generates one or more entity names (for example, "employee entity", "date entity", "overtime template", etc.). By associating one or more entity names with the initial template, the first template "Tell me the {overtime entity} of the {date entity} of the {employee entity}" may be generated. Further, the processing unit 30 generates one or more templates (for example, the second to fifth templates) having different positions or numbers of variables from the generated first template, and stores them in the template storage unit 102. You may.

Note that a template in which each variable is associated with an entity name may be considered as a type of conversational sentence information. In this case, the template storage unit 102 stores in advance an initial template in which the entity name does not correspond to each variable, and the conversation sentence information which is a template in which the entity name is associated with each variable is, for example, a conversation sentence. It may be stored in the information storage unit 14a.

The table storage unit 11a stores one or more table information. The one or more table information stored is usually the same information as the one or more table information stored in the table storage unit 11 in the first embodiment. The table information is not repeated as described above. Specifically, for example, the table information 1 to 3 shown in FIG. 4 may be stored in the table storage unit 11a.

Note that the table information may also have a table name, for example. The table name is the name of the table. The table name contains, for example, information that may form at least a portion of the action name. The information that can form at least a part of the action name is, for example, "employee", "working time", "business", and the like. Specifically, the table names included in the table information 1 to 3 are, for example, "employee master", "business master", "overtime time table", etc., but the format does not matter.

One or two or more intents are stored in the intent storage unit 12a. The intent will not be repeated as described above. The stored one or more intents are usually the same information as the one or more intents stored in the intent storage unit 12 in the first embodiment. One or more intents are generated by the intent generation unit 301 described later, and are stored in the intent storage unit 12a by the intent storage means 3061 described later. Specifically, for example, the

intents

1 and 2 shown in FIG. 5 may be stored in the intent storage unit 12a.

One or two or more API information is stored in the API information storage unit 13a. Since the API information has been described above, it will not be repeated. The one or more API information stored is usually the same information as the one or more API information stored in the API information storage unit 13 in the first embodiment. One or more API information is extracted from the API design document by the API information extraction means 3011 described later, and is stored in the API information storage unit 13a by the storage unit 306 described later. Specifically, for example, two or more API information shown in FIG. 6 may be stored in the API information storage unit 13a.

The conversation sentence information storage unit 14a stores one or more conversation sentence information in association with the intent. The conversational text information is not repeated as described above. The one or more conversational sentence information stored is usually the same information as the one or more conversational sentence information stored in the conversational sentence information storage unit 14 in the first embodiment. One or more conversational sentence information is generated by the conversational sentence information generation unit 303 described later, and is stored in the conversational sentence information storage unit 14a by the conversational sentence information storage means 3063 described later.

Specifically, in the conversation sentence information storage unit 14a, for example, as shown in FIG. 7, two or more conversation sentence information (here, five) are associated with the intent “overtime time inquiry intent”. Conversational sentence information 1 to 5) may be stored. Further, in the conversation sentence information storage unit 14a, two or more conversation sentence information (here,

conversation sentence information

1, 2, etc.) may be stored in association with the intent "working time data inquiry intent". ..

One or more entities are stored in the entity storage unit 15a in association with the entity name. The entity name and entity are not repeated as described above. The one or more entities stored in association with the entity name are usually the same information as the one or more entities stored in the entity storage unit 15 in association with the entity name in the first embodiment.

The entity name is generated by the entity name generation means 3021 described later, and one or more entities are generated by the entity generation means 3022 described later. The generated one or more entities are stored in the entity storage unit 15a by the entity storage means 3062 in association with the generated entity name.

Specifically, as shown in FIG. 8, for example, one or more entities "Ichiro Nakamura" and "Jiro Otsuka" are stored in the entity storage unit 15a in association with the entity name "employee entity". In addition, one or more entities "last month" and "yesterday" are stored in association with the entity name "date entity", and one or more entities "overtime hours" are associated with the entity name "overtime entity". "And" overtime "etc. may be stored. In addition, one or more entities corresponding to the "employee entity" may be a part of the name. A part of the name may be, for example, a surname such as "Nakamura" or "Otsuka", or an m character from the front (for example, "Nakamura Ichi" or "Otsuka Ni" which is three characters from the front).

One or two or more entity mapping information is stored in the entity mapping information storage unit 16a. The entity mapping information will not be repeated as described above. The one or more entity mapping information stored is usually the same information as the one or more entity mapping information stored in the entity mapping information storage unit 16 in the first embodiment. One or more entity mapping information is generated by the entity mapping information generation unit 304 described later, and is stored in the entity mapping information storage unit 16a by the entity mapping information storage means 3064 described later.

Specifically, the entity mapping information storage unit 16a may store one or more entity mapping information (for example, entity mapping information 1 to 4, etc.) as shown in FIG. 9, for example.

One or two or more PK items are stored in the PK item storage unit 17a. The PK items are not repeated as described above. The one or more PK items stored are usually the same information as the one or more PK items stored in the PK item storage unit 17 in the first embodiment. One or more PK items are generated by the PK item generation unit 305 described later, and are stored in the PK item storage unit 17a by the PK item storage means 3065 described later.

Specifically, the PK item storage unit 17a may store, for example, two or more PK items (for example, PK items 1 to 4, etc.) as shown in FIG.

The reception unit 20 receives various types of information. The various types of information include, for example, API design documents and templates. The reception unit 20 receives information such as an API design document from a terminal, for example, but may receive information via an input device such as a keyboard or a touch panel. Alternatively, the reception unit 20 may accept information read from a recording medium such as a disk or a semiconductor memory, and the mode of acceptance is not limited.

The processing unit 30 performs various processes. The various processes include, for example, intent generation unit 301, API information extraction means 3011, action name acquisition means 3012, intent generation means 3013, entity generation unit 302, entity name generation means 3021, entity generation means 3022, and conversational text. Information generation unit 303, entity acquisition means 3031, conversation sentence generation means 3032, entity mapping information generation unit 304, PK item generation unit 305, storage unit 306, intent storage means 3061, entity storage means 3062, conversation sentence information storage means 3063. , Entity mapping information storage means 3064, PK item storage means 3065, and the like.

Note that the processing unit 30 also performs processing for accumulating information such as API design documents received by the reception unit 20 in the document storage unit 101 or the like.

Further, when the table contained in the table information stored in the table storage unit 11 is, for example, a document file such as a word and is not converted into a tabular format, the processing unit 30 performs the document file. A process of converting the document file into a tabular format may be performed based on the delimiter provided in the character string of. The process of converting a document file such as a word into a tabular format is known, and the description thereof will be omitted.

Further, the processing unit 30 may perform a process of generating the first template by associating one or more entity names generated by the entity generation unit 302 described later with the initial template. For example, if the initial template is "Tell me {} of {} of {}" and the three entity names "employee entity", "date entity", and "overtime entity" are acquired, the processing unit 30 , The three entity names may be associated with the initial template to generate the first template "Tell me the {overtime entity} of the {date entity} of the {employee entity}".

Further, the processing unit 30 generates one or more templates (for example, second to fifth templates) having different positions or numbers of variables from the generated first template, and stores them in the template storage unit 102. You may.

Specifically, the processing unit 30 generates a second template by exchanging the first variable and the second variable with respect to the first template, for example. Further, the template generation unit generates a third template by deleting the second variable and the particle "no" following the first template, for example. Further, the template generation unit generates a fourth template by deleting the first variable and the subsequent particle "no" from the first template, for example. Further, the template generation unit deletes the first variable and the subsequent particle "no" and the second variable and the subsequent particle "no" from the first template, for example, so that the fifth template can be used. Generate a template for. Then, the processing unit 30 associates the generated first to fifth templates with, for example, the action name acquired by the action name acquisition means 3012 or the intent name generated by the intent generation means 3013, and the template storage unit 30. Accumulate in 102.

The intent generation unit 301 acquires an action name using the first document stored in the document storage unit 101, and generates an intent having the acquired action name. The generated intent may be the same as the acquired action name. That is, the intent may be generated, for example, by acquiring the action name.

As mentioned above, the first document has an action name or a part of the action name. The intent generation unit 301 acquires an action name from the first document having the action name, and generates an intent having the acquired action name, for example. Alternatively, the intent generation unit 301 may acquire a part of the action name from the first document, modify the part to form the action name, and acquire the intent having the action name.

The transformation is, for example, the addition of a predetermined character string to the part, but the conversion of the character string constituting the part to another type of character string may be performed, and the mode thereof does not matter. The addition is, for example, an addition to the end of the character string constituting the part, but may be an addition to the beginning of the character string, an addition to the inside of the character string, and the position thereof does not matter. The conversion includes, for example, conversion from a kanji character string to a kana character string, conversion from a kanji / kana mixed character string to an alphabetic character string, and the like, but the type does not matter.

For example, the action name dictionary is stored in the storage unit 10. The action name dictionary is a dictionary related to action names. For example, one or more action names are registered in the action name dictionary. Alternatively, one or more pairs of action names and one or more vocabularies may be registered in the action name dictionary. The action name dictionary may be, for example, the intent name dictionary described in Specific Example 2.

The intent generation unit 301 performs morphological analysis on the first document, for example, and acquires one or more words. Next, the intent generation unit 301 includes the word and one or more action names (or one or more corresponding to the action name) registered in the action name dictionary for each one or more acquired words. Get the similarity with each vocabulary). Next, the intent generation unit 301 performs the word and action corresponding to the highest similarity among the one or more similarity acquired for all combinations of one or more words and one or more action names and the like. Specify the combination with the name etc. Then, the intent generation unit 301 generates an intent having a word, an action name, or the like that constitutes the specified combination.

The intent generation unit 301 acquires an action name from the API design document stored in the document storage unit 101, for example, by the above procedure, and generates an intent having the acquired action name. For example, when the stored API design document has the action name "overtime hours inquiry", the intent generation unit 301 acquires the action name "overtime hours inquiry" from the API design document and obtains the action name "overtime hours inquiry". By adding "inquiry" to the end, an intent "overtime hour inquiry intent" having the action name "overtime hour inquiry" may be generated.

Or, for example, when the stored API design document contains one or more API information, and each of the one or more API information includes information that can form a part of the action name, an intent is generated. The API information extracting means 3011 constituting the unit 301 extracts one or more API information from the API design document. The information that can form a part of the action name is the API information included in the API design document, for example, "search_overtime hours (employee code, start date, end date)" or "search_work" shown in FIG. In the case of an API statement such as "time (employee code, work date, business ID)", it is a character string such as "overtime hours" or "working hours" sandwiched between "search" and "(".

The action name acquisition means 3012 acquires information that can form a part of the action name from the API information for each of the extracted one or more API information. For example, from the API information "search_overtime hours (employee code, start date, end date)", the character string "overtime hours" sandwiched between "search" and "(" is acquired. Action name acquisition means 3012 Acquires the action name "overtime hours inquiry" by adding "inquiry" to the end of the information acquired in this way.

The intent generation means 3013 generates an intent having the action name by adding "intent" to the end of each one or more action names acquired by the action name acquisition means 3012. For example, the intent generation means 3013 adds "intent" to the end of the acquired action name "overtime hour inquiry" to add an intent "overtime hour inquiry intent" having the action name. Generate. Similarly, from the action name "working time data inquiry", an intent "working time data inquiry intent" having the action name is generated.

The entity generation unit 302 generates an entity name which is an attribute name by using, for example, a second document stored in the document storage unit 101. The entity name may be generated, for example, by acquiring the attribute name from the second document.

Specifically, the entity generation unit 302 may acquire the attribute name from the second document having the attribute name and generate the entity name which is the acquired attribute name, for example. Alternatively, the entity generation unit 302 acquires the attribute name and one or more words corresponding to the attribute name from the second document having the attribute name and one or more words corresponding to the attribute name, and the acquired attribute. You may generate an entity name that is a name and one or more entities that are one or more acquired words.

Alternatively, the entity generation unit 302 may generate an entity name which is an attribute name by using, for example, one or more table information stored in the table storage unit 11a.

For example, the attribute name dictionary is stored in the storage unit 10. The attribute name dictionary is a dictionary related to attribute names. For example, one or more attribute names are registered in the attribute name dictionary. Alternatively, one or more pairs of attribute names and one or more vocabularies may be registered in the attribute name dictionary. The attribute name dictionary may be, for example, the entity name dictionary described in Specific Example 2.

The entity generation unit 302 performs morphological analysis on the second document, for example, and acquires one or more words. Next, the entity generation unit 302 includes the word and one or more attribute names (or one or more corresponding to the attribute name) registered in the attribute name dictionary for each of the acquired one or more words. Get the similarity with each vocabulary). Next, the entity generation unit 302 performs the word and attribute name corresponding to the highest similarity among the one or more similarity acquired for all combinations of one or more words and one or more entity names and the like. Specify the combination with etc. Then, the entity generation unit 302 generates an entity name having a word, an attribute name, or the like that constitutes the specified combination.

For example, the entity generation unit 302 corresponds to the attribute name and the attribute name from the table of the table information for each one or more table information stored in the table storage unit 11a by the above procedure. The above words are acquired, and an entity name that is an acquired attribute name and one or more entities that are one or more acquired words are generated.

For example, from a table (not shown) in which the attribute name "employee" and one or more words "Ichiro Nakamura", "Jiro Otsuka", etc. corresponding to the attribute name are registered, the attribute name "employee" and the relevant 1 The above words "Ichiro Nakamura", "Jiro Otsuka", etc. are acquired, the acquired entity name "employee", and one or more acquired words "Ichiro Nakamura", " Jiro Otsuka "etc. may be generated.

Note that the entity name that is the attribute name includes, for example, the case where the entity name is an attribute name that includes only a part of the acquired attribute name. That is, the entity generation unit 302 acquires an attribute name and one or more words corresponding to the attribute name from the table of the table information for each one or more table information, and is a part of the acquired attribute name. You may generate an entity name and one or more entities that are one or more acquired words. For example, from a table (not shown) in which the attribute name "employee name" and one or more words "Ichiro Nakamura", "Jiro Otsuka" corresponding to the attribute name are registered, the attribute name "employee name" and the attribute name "employee name" The one or more words "Ichiro Nakamura", "Jiro Otsuka", etc. are acquired, and the entity name "employee" that is a part of the acquired attribute name, and one or more entities that are one or more acquired words " "Ichiro Nakamura", "Jiro Otsuka", etc. may be generated.

Note that the generated entity may also be a part of the acquired entity. For example, when the word "Ichiro Nakamura" is acquired, the entity generation unit 302 may generate the entity "Nakamura" containing the first two characters of the word, the entity "Nakamura Ichi" including the preceding three characters, and the like. Good.

Further, the entity name which is the acquired attribute name includes the case where the entity name is, for example, an attribute name obtained by converting at least a part of the acquired attribute name. The conversion is not limited to conversion to a different language (for example, translation), but may be conversion from alphabetical notation to kanji display (for example, conversion from "SHAIN" to "employee"). .. That is, the entity generation unit 302 acquires an attribute name and one or more words corresponding to the attribute name from the table possessed by the table information for each one or more table information, and at least a part of the acquired attribute name is acquired. An entity name that is a converted attribute name and one or more entities that are one or more acquired words may be generated.

For example, from the table information 1 shown in FIG. 4, the attribute name “SHAIN_NAME” and one or more words “Ichiro Nakamura”, “Jiro Otsuka” corresponding to the attribute name are acquired, and a part of the attribute name is obtained. An entity name "employee entity" including the attribute name "employee" converted from "SHAIN", and one or more entities "Ichiro Nakamura", "Jiro Otsuka", etc., which are one or more acquired words, may be generated. ..

Specifically, for example, the entity name generation means 3021 constituting the entity generation unit 302 acquires an attribute name from the table information for each one or more table information, and generates an entity name which is the attribute name. The generated entity name is stored in the entity storage unit 15a by, for example, the entity storage means 3062 described later.

For example, the entity generation means 3022 acquires one or more words corresponding to the acquired attribute names from the table corresponding to the entity name for each one or more entity names generated by the entity name generation means 3021. Generate one or more entities that are the one or more words. The generated one or more entities are stored in the entity storage unit 15a in association with the entity name by, for example, the entity storage means 3062 described later.

For example, for table information 1, the entity name generation means 3021 acquires the attribute name "SHAIN_NAME" from the table information 1, and converts "SHAIN", which is a part of the attribute name, into an "employee". Generate the entity name "employee entity" with "" added. The entity storage means 3062 stores the generated entity name “employee entity” in the entity storage unit 15a.

The entity generation means 3022 selects one or more words "Ichiro Nakamura", "Jiro Otsuka", etc. corresponding to the attribute name from the table (that is, the employee master) corresponding to the generated entity name "employee entity". Acquire and generate one or more entities "Nakamura", "Otsuka", etc. that are a part of them (for example, the first two characters). The entity storage means 3062 stores one or more generated entities "Nakamura", "Otsuka", etc. in the entity storage unit 15a in association with the entity name "employee entity".

Further, for the table information 2, the entity name generating means 3021 acquires the attribute name "GYOUMU_NAME", and the entity name "business entity" is obtained by adding "entity" to the "business" obtained by partially converting "GYOUMU". To generate. The generated entity name "business entity" is stored in the entity storage unit 15a. The entity generation means 3022 acquires "A business", "B business", etc. corresponding to the attribute name from the table (that is, business master) corresponding to the generated entity name "business entity", and has one or more. Generate entities "A business", "B business", etc. The one or more generated entities "A business", "B business", etc. are stored in the entity storage unit 15a in association with the entity name "business entity".

For the table information 3, three attribute names "ZIKANGAI_TIME", "TOTAL_TIME", and "TIME_CARD_DATE" are acquired, and a part "ZIKANGAI" of the first attribute name "ZIKANGAI_TIME" is converted to "ZIKANGAI". The entity name "overtime entity" is generated by adding "entity" to "overtime". In addition, an entity name "working time entity" is generated by adding "entity" to "working hours" obtained by converting the second attribute name "TOTAL_TIME", and a part of the third attribute name "TIME_CARD_DATE". An entity name "date entity" is generated by adding "entity" to "date" converted from "DATE". The three generated entity names "overtime entity", "working time entity", and "date entity" are stored in the entity storage unit 15a.

Next, for the entity name "overtime entity", the entity "overtime hours" in which the first attribute name "ZIKANGAI_TIME" is converted, and the entity "overtime" in which a part thereof is converted are generated. Further, for the entity name "working time entity", the entity "working time" obtained by converting the second attribute name "TOTAL_TIME" and the entity "working" obtained by converting a part thereof are generated.

For the entity name "date entity", for example, a set of "date" and one or more corresponding words "last month", "yesterday", etc. are stored in the storage unit 10. The entity generation means 3022 acquires one or more words "last month", "yesterday", etc. corresponding to the word "date" included in the "date entity" from the storage unit 10, and the acquired word entity "last month", Generate "yesterday" etc.

Then, one or more entities "overtime hours", "overtime", etc. are accumulated in association with the entity name "overtime entity", and one or more entities "working hours" are associated with the entity name "working hours entity". , "Work" etc. are accumulated, and one or more entities "last month", "yesterday" etc. are accumulated in association with the entity name "date entity".

As a result, the entity storage unit 15a stores, for example, an entity name and a set of one or more sets of entities corresponding to the entity name, as shown in FIG.

The conversational sentence information generation unit 303 inserts one or more of the one or more entities generated by the entity generation unit 302 into the variables of the template stored in the template storage unit 102, thereby causing one or more. Generates conversational text information.

When the template has two or more variables, for example, the conversation sentence information generation unit 303 acquires one entity from one or more entities generated by the entity generation unit 302 for each of the two or more table information, and the acquired entity is obtained. One or more conversational sentence information may be generated by sequentially inserting two or more entities into two or more variables of one template stored in the template storage unit 102.

Specifically, for example, the initial template "{} {} {} of {}" having three variables is stored in the template storage unit 102, and the table information 1 to 3 are stored in the table storage unit 11a. When stored, the conversation sentence information generation unit 303 acquires one entity "Ichiro Nakamura" among one or more entities "Ichiro Nakamura", "Jiro Otsuka", etc. generated from the table information 1, and table information. One or more entities "last month", "yesterday", etc. generated from 2 get one entity "last month", and one or more entities "overtime hours", "overtime", etc. generated from table information 3 Get one of the entities "overtime hours". Then, the conversation sentence information generation unit 303 sets the acquired three entities "Ichiro Nakamura", "last month", and "overtime hours" into the three variables (first variable, second variable, and 3) of the above template. By inserting in order in the second variable), the conversational sentence information "Tell me the overtime hours of last month of Ichiro Nakamura" is acquired.

For example, an ID (for example, "1", "2", etc.) is associated with each of two or more variables of the initial template, and the storage unit 10 has an ID of the variable and an ID of table information (for example, "2"). , "1", "2", etc.) and correspondence information indicating correspondence may be stored. In this case, the conversational sentence information generation unit 303 refers to the corresponding information, and the entity concerned obtains each of the two or more entities acquired from the two or more table information (for example, table information 1, table information 2, etc.). You can insert it in the variable corresponding to the acquired table information. As a result, even if the order of the two or more variables possessed by the template and the order of the two or more table information stored in the table storage unit 11a do not correspond in advance, two or more acquired two or more entities can be obtained. Can be accurately inserted into the variable of.

In addition, the conversation sentence information generation unit 303 inserts the acquired three entities "Ichiro Nakamura", "last month", and "overtime hours" into the three variables of the above template by changing the order of the previous two. , Generate conversational text information "Tell me about the overtime hours of Ichiro Nakamura last month".

In this way, based on one template "Tell me {} of {} of {}" and table information 1 to 3, two conversational sentence information "Tell me the overtime hours of Ichiro Nakamura last month", and " "Tell me about the overtime hours of Ichiro Nakamura last month" is generated.

Note that the entity inserted in the variable may be a part of the acquired entity. For example, for the entity "Ichiro Nakamura" acquired from the table information 1, the preceding two characters "Nakamura" or the preceding three characters "Nakamura Ichi" may be inserted.

The entity acquisition means 3031 constituting the conversational sentence information generation unit 303 corresponds to the entity name associated with each of the one or more variables of the template for each of the two or more templates stored in the table storage unit 11a. Acquire one entity to be executed from the entity storage unit 15a.

Specifically, for example, the first to fifth templates as shown in FIG. 14 are stored in the template storage unit 102 in association with the action name “overtime hours inquiry”, and are stored in the table storage unit 11a as described above. When the table information 1 to 3 is stored, the entity acquisition means 3031 has one or more entities "Nakamura" corresponding to the entity name "employee entity" associated with the first variable with respect to the first template. Acquire one entity "Ichiro Nakamura" from "Ichiro", "Jiro Otsuka", etc., and one or more entities "last month", "yesterday" corresponding to the entity name "date entity" associated with the second variable. One or more entities "overtime hours", "overtime", etc. that correspond to the entity name "overtime entity" associated with the third variable by acquiring one entity "last month", etc. Get "overtime hours".

Further, the entity acquisition means 3031 is one of one or more entities "last month", "yesterday", etc. corresponding to the entity name "date entity" associated with the first variable with respect to the second template. Acquire the entity "last month" and select one entity "Ichiro Nakamura" from one or more entities "Ichiro Nakamura", "Jiro Otsuka", etc. corresponding to the entity name "employee entity" associated with the second variable. Acquires and acquires one entity "overtime hours" such as one or more entities "overtime hours" and "overtime work" corresponding to the entity name "overtime work entity" associated with the third variable.

Similarly, for the third template, "Ichiro Nakamura" among one or more entities "Ichiro Nakamura", "Jiro Otsuka", etc. corresponding to the entity name "employee entity" associated with the first variable. Is acquired, and "overtime hours" among one or more entities "overtime hours", "overtime hours", etc. corresponding to the entity name "overtime work entity" associated with the second variable is acquired.

Also, for the fourth template, "overtime hours" out of one or more entities "overtime hours", "overtime hours", etc. corresponding to the entity name "overtime entity" associated with the first variable is acquired. Then, "Ichiro Nakamura" is acquired from one or more entities "Ichiro Nakamura", "Jiro Otsuka", etc. corresponding to the entity name "employee entity" associated with the second variable.

Further, for the fifth template, "overtime hours" among one or more entities "overtime hours", "overtime hours", etc. corresponding to the entity name "overtime work entity" associated with the first variable is acquired. Will be done.

Note that the entity acquisition means 3031 may acquire one or more entities corresponding to one or more entity names by using, for example, the entity mapping information generated by the entity mapping information generation unit 304 described later. That is, the entity acquisition means 3031 refers to the entity mapping information having the entity name for each one or more entity names, and from the table identified by the table identifier having the entity mapping information, the entity mapping is performed. One or more entities corresponding to the translation item name of the information may be acquired.

Specifically, for example, for the entity name "employee entity", the entity acquisition means 3031 refers to the entity mapping information 1 including the entity name and identifies it by the table identifier "MST_SHAIN" possessed by the entity mapping information 1. One or more entities "Ichiro Nakamura", "Jiro Otsuka", etc. corresponding to the translation item name "SHAIN_NAME" possessed by the entity mapping information 1 may be acquired from the table (that is, the business master).

The conversation sentence generation means 3032 inserts the entity acquired by the entity acquisition means 3031 at the position of one or more variables of each of the two or more templates stored in the template storage unit 102. Generate one or more conversational sentence information.

Specifically, for example, in contrast to the first template "Tell me the {overtime entity} of the {date entity} of the {employee entity}", the conversation sentence generation means 3032 has three of the first template. The preceding two characters "Nakamura" of the entity "Ichiro Nakamura" acquired by the entity acquisition means 3031 are inserted at the position of the first variable among the variables. Further, the conversation sentence generation means 3032 inserts the acquired entity "last month" at the position of the second variable, and further inserts the acquired entity "overtime hours" at the position of the third variable. As a result, the first conversational sentence information "Tell me about Nakamura's overtime hours last month" is generated.

Also, in contrast to the second template "Tell me the {overtime entity} of the {date entity} of the {employee entity}", the conversation sentence generation means 3032 has the acquired entity "at the position of the first variable." Insert "Last month", insert the previous two characters "Nakamura" of the acquired entity "Ichiro Nakamura" at the position of the second variable, and further insert the acquired entity "Overtime hours" at the position of the third variable. By inserting "", the second conversational text information "Tell me about Nakamura's overtime hours last month" is generated.

Similarly, for the third template "Tell me the {overtime entity} of the {employee entity}", the third conversational text information "Tell me the overtime hours of Nakamura" is generated, and the fourth template "{ For "Tell me {overtime entity} of the date entity}", the fourth conversation information "Tell me the overtime hours of last month" is generated, and for the fifth template "Tell me {overtime entity}" Then, the fifth conversational sentence information "Tell me about overtime hours" is generated.

Further, the conversation sentence generation means 3032 may also generate one or more entity information for each one or more generated conversation sentence information, for example. Since the entity information has been described in the first embodiment, it will not be repeated.

Specifically, the conversation sentence generation means 3032 extracts one or more entities possessed by one conversation sentence information, and acquires a start position, an end position, and an entity name for each of the extracted one or more entities. Then, one or more entity information having an entity, a start position and an end position, and an entity name is generated. In addition, one or more entity information generated for one conversational sentence information is associated with the one conversational sentence information by, for example, the conversational sentence information storage means 3063 described later, and is stored in the conversational sentence information storage unit 14a. It may be accumulated.

Specifically, for example, for the first conversational sentence information "Tell me about Nakamura's overtime hours last month", the conversational sentence generation means 3032 has three entities "Nakamura", "Last month", and "Overtime hours" is extracted, and the start position "1" and the end position "2" of the first entity "Nakamura" and the entity name "employee entity" are acquired. Further, the conversation sentence generation means 3032 acquires the start position "4" and the end position "5" of the second entity "last month" and the entity name "date entity", and further, the third entity "overtime work". The start position "7" and the end position "10" of "time" and the entity name "overtime entity" are acquired.

Then, the conversation sentence generation means 3032 has the entity "Nakamura", the start position "1", the end position "2", and the first entity information "Nakamura, 1, 2, employee" having the entity name "employee entity". Create an "entity". Further, the conversation sentence generation means 3032 has a second entity information "last month, 4, 5, date" having an entity "last month", a start position "4", an end position "5", and an entity name "date entity". Create an "entity". Further, the conversation sentence generation means 3032 has a third entity information “overtime hours, 7, 10” having an entity “overtime hours”, a start position “7”, an end position “20”, and an entity name “overtime entity”. "Generate. The first to third entity information generated in this way is associated with the first conversational sentence information "Tell me Nakamura's overtime hours last month" by, for example, the conversational sentence information storage means 3063 described later, and the conversational sentence. It is stored in the information storage unit 14a.

In response to the second conversational sentence information "Tell me about Nakamura's overtime hours last month", the conversational sentence generation means 3032 extracts three entities "last month", "Nakamura", and "overtime hours". The start position "1" and the end position "2" of the first entity "last month" and the entity name "date entity" are acquired. Further, the conversation sentence generation means 3032 acquires the start position "4" and the end position "5" of the second entity "Nakamura" and the entity name "employee entity", and further, the third entity "overtime work". The start position "7" and the end position "10" of "time" and the entity name "overtime entity" are acquired.

Then, the conversation sentence generation means 3032 uses the first entity information "last month, 1, 2, date entity", the second entity information "Nakamura, 4, 5, employee entity", and the third entity information "overtime hours". , 7, 10, Overtime entity "is generated. The generated first to third entity information is stored in the conversation text information storage unit 14a in association with the second conversation text information "Tell me the overtime hours of Nakamura last month".

Similarly, for the third conversational text information "Tell me about Nakamura's overtime hours", the first entity information "Nakamura, 1, 2, employee entity" and the second entity information "Overtime hours, 4," 7. Overtime entity ”is acquired, and the acquired first and second entity information is stored in the conversation sentence information storage unit 14a in association with the third conversation sentence information. In addition, for the fourth conversational text information "Tell me about overtime hours last month", the first entity information "Last month, 1, 2, date entity" and the second entity information "Overtime hours, 4, 7" , Overtime entity ”is acquired, and the acquired first and second entity information is stored in the conversation text information storage unit 14a in association with the fourth conversation text information. Furthermore, the first entity information "overtime hours, 1, 4, overtime entities" is acquired for the fifth conversational sentence information "tell me overtime hours", and the acquired first entity information is the fifth. It is stored in the conversation sentence information storage unit 14a in association with the conversation sentence information of.

For the intent "overtime work time inquiry", for example, 16 conversation sentence information 1 to 15 are generated, and 4 entity information "Ichiro Nakamura", "1, for the conversation sentence information 1". 4, Employee entity "," Yesterday, 6, 7, Date entity "," A matter, 9, 11, Business entity ", and" Working hours, 13, 16, Working hours entity "are generated. Further, one or more entity information is generated for the conversational sentence information 2 to 15.

In this way, in the conversation sentence information storage unit 14a, for example, as shown in FIG. 7, five conversation sentence information 1 to 5 are stored in association with the intent "overtime time inquiry intent", and further, the said One or more entity information is stored in association with each of the five conversational sentence information 1 to 5. In addition, 16 conversational sentence information 1 to 15 are stored in association with the intent "working time data inquiry intent", and one or more conversational sentence information 1 to 15 are associated with each of the 16 conversational sentence information 1 to 15. Entity information is stored.

The entity mapping information generation unit 304 is a table identifier and an attribute name in the table identified by the table identifier from the table information for each one or more table information stored in the table storage unit 11a. The translation item name which is the information for identifying the entity to be converted and the one entity corresponding to the translation item are acquired, and the entity name corresponding to one entity is acquired from the entity storage unit 15a, and the table identifier is obtained. And generate entity mapping information with translation item name and entity name. The generated one or more entity mapping information is stored in the entity mapping information storage unit 16a by, for example, the entity mapping information storage means 3064 described later.

Specifically, for example, for the table information 1, the entity mapping information generation unit 304 has the table identifier "MST_SHAIN" from the table information 1 and the table (that is, the employee master) identified by the table identifier. The translation item name "SHAIN_NAME", which is the attribute name of the above and is the information that identifies the entity to be converted (that is, "Ichiro Nakamura", "Jiro Otsuka", etc.), and the one entity "Ichiro Nakamura" corresponding to the translation item. And, the entity name "employee entity" corresponding to the one entity is acquired from the entity storage unit 15a. Then, the entity mapping information generation unit 304 generates the entity mapping information 1 having the acquired table identifier "MST_SHAIN", the acquired translation item name "SHAIN_NAME", and the acquired entity name "employee entity".

Further, with respect to the table information 2, the entity mapping information generation unit 304 has the table identifier "MST_GYOUMU" from the table information 2 and the attribute name in the table (that is, the business master) identified by the table identifier. , Acquire the translation item name "GYOUMU_NAME" which is the information that identifies the entity to be converted (that is, "A matter", "B matter", etc.) and one entity "A matter" corresponding to the translation item. At the same time, the entity name "business entity" corresponding to the one entity is acquired from the entity storage unit 15a. Then, the entity mapping information generation unit 304 generates the entity mapping information 4 having the table identifier "MST_GYOUMU", the translation item name "GYOUMU_NAME", and the entity name "business entity".

For the table information 3, the entity mapping information 3 having the table identifier "TRN_TIME_CARD", the translation item name "TIME_CARD_DATE" and the entity name "date entity", and the table identifier "TRN_TIME_CARD" and the translation item name "ZIKANGAI_TIME" Entity mapping information 3 having the entity name "overtime entity" is generated.

The entity mapping information 1 to 4 and the like thus generated are stored in the entity mapping information storage unit 16a by the entity mapping information storage means 3064. As a result, for example, the entity mapping information 1 to 4 shown in FIG. 9 is stored in the entity mapping information storage unit 16a.

The PK item generation unit 305 specifies a table identifier and a primary key of the table identified by the table identifier from the table information for each one or more table information stored in the table storage unit 11a. The identifier is acquired, and a PK item which is information having the acquired table identifier and the acquired primary key identifier is generated.

Specifically, for example, with respect to the table information 1, the PK item generation unit 305 uses the table identifier "MST_SHAIN" from the table information 1 and the main of the table (that is, the employee master) identified by the table identifier. The key identifier "SHAIN_CODE" is acquired, and PK item 1 which is information having the table identifier "MST_SHAIN" and the primary key identifier "SHAIN_CODE" is generated.

Further, with respect to the table information 2, the PK item generation unit 305 uses the table identifier "MST_GYOUMU" and the primary key identifier "GYOUMU_ID" of the table (that is, the business master) identified by the table identifier. To generate PK item 4, which is information having the table identifier "MST_GYOUMU" and the primary key identifier "GYOUMU_ID".

For the table information 3, a PK item 2 having a table identifier "TRN_TIME_CARD" and a primary key identifier "SHAIN_CODE" and a PK item 3 having a table identifier "TRN_TIME_CARD" and a primary key identifier "TIME_CARD_DATE" are generated. Will be done.

The PK items 1 to 4 and the like thus generated are stored in the PK item storage unit 17a by the PK item storage means 3065. As a result, for example, the PK items 1 to 4 shown in FIG. 10 are stored in the PK item storage unit 17a.

For example, the storage unit 306 stores the intent generated by the intent generation unit 301 in the intent storage unit 12a, stores the entity name generated by the entity generation unit 302 in the entity storage unit 15a, and stores conversational text information. One or more conversational sentence information generated by the generation unit 303 is associated with the intent and stored in the conversational sentence information storage unit 14a. It should be noted that associating the conversational sentence information with the intent includes, for example, the case where the conversational sentence information has an intent.

Alternatively, the storage unit 306 stores the intent generated by the intent generation unit 301 in the intent storage unit 12a, and associates one or more entities with one or more entity names generated by the entity generation unit 302. One or more conversational sentence information stored in the entity storage unit 15a and generated by the conversational sentence information generation unit 303 may be stored in the conversational sentence information storage unit 14a in association with the intent.

The intent storage means 3061 constituting the storage unit 306 stores one or more intents generated by the intent generation means 3013 in the intent storage unit 12a.

The entity storage means 3062 is one or more entities generated by the entity generation means 3022 in association with the entity name for each one or more entity names generated by the entity name generation means 3021, and corresponds to the entity name. One or more entities that are one or more words acquired from the table to be stored are stored in the entity storage unit 15a.

The conversational sentence information storage means 3063 associates one or more conversational sentence information generated by the conversational sentence generating means 3032 with the intent for each one or more intents generated by the intent generation means 3013. It accumulates in the storage unit 14a.

Further, the conversation sentence information storage means 3063 associates one or more entity information generated by the conversation sentence generation means 3032 with the conversation sentence information for each one or more conversation sentence information generated by the conversation sentence generation means 3032, for example. It may be stored in the conversation text information storage unit 14a.

The entity mapping information storage unit 3064 stores one or more entity mapping information generated by the entity mapping information generation unit 304 in the entity mapping information storage unit 16a.

The PK item storage unit 3065 stores one or more PK items generated by the PK item generation unit 305 in the PK item storage unit 17a.

The output unit 40 outputs various information. The various types of information include, for example, intents, entities, conversational text information, entity mapping information, PK items, and the like. The output unit 40 transmits information such as an intent to the information output device 500 described in the first embodiment, but may output information via an output device such as a display or a speaker.

The output unit 40 is stored in, for example, one or more intents stored in the intent storage unit 12a, one or more conversational sentence information stored in the conversational sentence information storage unit 14a, and an entity storage unit 15a. One or more entities, one or more entity mapping information stored in the entity mapping information storage unit 16a, and one or more PK items stored in the PK item storage unit 17a are transmitted to the information output device 500.

In the information output device 500, for example, the reception unit 2 receives the information such as one or more intents, and the processing unit 3 receives the received information such as one or more intents in the intent storage unit 12. It also performs processing to accumulate in the like.

Next, the operation of the information generation device 100 will be described with reference to the flowchart of FIG. FIG. 15 is a flowchart illustrating the operation of the information generation device 100.

(Step S1501) The intent generation unit 301 acquires an action name from the first document stored in the document storage unit 101, and generates an intent having the action name.

(Step S1502) The storage unit 306 stores the intent generated in step 1501 in the intent storage unit 12a.

(Step S1503) The entity generation unit 302 acquires an attribute name from the second document stored in the document storage unit 101, and generates an entity name which is the acquired attribute name. As described above, the second document may be the same document as the first document.

(Step S1504) The storage unit 306 stores the entity name generated in step S1503 in the entity storage unit 15a.

(Step S1505) The processing unit 30 generates a template having a variable associated with the entity name by associating the entity name generated in step 1503 with the variable of the initial template.

(Step S1506) The storage unit 306 stores the template generated in step S1505 in the conversation text information storage unit 14a in association with the intent accumulated in step S1502. After that, the process ends.

In the flowchart of FIG. 15, the initial template has two or more variables. In step S1503, the entity generation unit 302 generates one or more entities, and in step S1505, the processing unit 30 steps. The intent generated in 1501 is associated with the last variable among the two or more variables of the initial template, and the one or more entity names generated in step 1503 are associated with the last of the two or more variables. By associating with one or more variables excluding variables, a template having one or more variables associated with entity names and variables associated with intents may be generated.

Further, in the flowchart of FIG. 15, steps S1503 and S1506 may be replaced with the following steps S1503a and S1506a.

(Step S1503a) The entity generation unit 302 acquires an attribute name and one or more words corresponding to the attribute name from the second document stored in the document storage unit 101, and the entity name which is the acquired attribute name, And generate one or more entities that are one or more acquired words.

(Step S1506a) The storage unit 306 stores the entity name generated in step S1503a and a set of one or more entities in the entity storage unit 15a.

Further, the following two steps S1507 and S1508 may be added after step S1506a.

(Step S1507) The conversational sentence information generation unit 303 inserts one or more of the one or more entities generated in step S1503a into the variables associated with the entity names in the template accumulated in step S1506. By doing so, one or more conversational sentence information is generated.

(Step S1508) The storage unit 306 stores one or more conversational sentence information generated in step S1507 in the conversational sentence information storage unit 14a in association with the intent accumulated in step S1502.

FIG. 16 is a flowchart illustrating an operation example of the information generation device 100 when the first document is an API design document and the second document is table information. Note that this flowchart is a process for one action name among one or more action names included in the API design document, and the same process is performed for other action names.

(Step S1601) The intent generation unit 301 acquires an action name from the API design document stored in the API information storage unit 13a, and generates an intent having the action name.

(Step S1602) The intent storage means 3061 stores the intent generated in step 1501 in the intent storage unit 12a.

(Step S1603) The entity generation unit 302 sets the initial value "1" in the variable i. The variable i in this flowchart is a variable for sequentially selecting unselected table information from one or more table information stored in the table storage unit 11a.

(Step S1604) The entity generation unit 302 determines whether or not there is the i-th table information. If there is the i-th table information, the process proceeds to step S1605. If not, the process proceeds to step S1608.

(Step S1605) The entity generation unit 302 acquires the attribute name and one or more words corresponding to the attribute name from the table of the i-th table information, the entity name which is the acquired attribute name, and the acquired 1 Generate one or more entities that are the above words.

(Step S1606) The entity storage means 3062 stores one or more entities generated in step S1605 in the entity storage unit 15a in association with the entity name stored in step S1602.

(Step S1607) The entity generation unit 302 increments the variable i. After that, the process returns to step S1604.

(Step S1608) The processing unit 30 associates one or more entity names generated by one or more executions of steps S1604 to S1607 with one or more variables of the initial template, and makes the number or position of the variables different. Generate one or more templates.

(Step S1609) The conversation sentence information generation unit 303 sets the initial value "1" in the variable j. The variable j in this flowchart is a variable for sequentially selecting an unselected template from the one or more templates generated in step S1608.

(Step S1610) The conversation sentence information generation unit 303 determines whether or not there is a j-th template. If there is a j-th template, the process proceeds to step S1611, otherwise the process proceeds to step S1613.

(Step S1611) The conversational sentence information generation unit 303 acquires one word from each of one or more tables stored in the table storage unit 11a and inserts it at the position of one or more variables of the j-th template. Generates conversational sentence information.

(Step S1612) The conversational sentence information generation unit 303 increments the variable j. After that, the process returns to step S1610.

(Step S1613) The conversation sentence information generation unit 303 associates one or more conversation sentence information generated by one or more executions of steps S1610 to S1612 with the intent accumulated in step S1602 and stores the conversation sentence information. Accumulate in part 14a. After that, the process ends.

Note that one of the specific operation examples of the information generation device 100 (hereinafter, specific example 1) is described in the explanation of the block diagram and will not be repeated. In Specific Example 1, an intent, a set of an entity name and one or more entities, and an example sentence of a conversation sentence are used using an API design document which is a kind of the first document and table information which is a kind of the second document. One or more conversational sentence information and the like are generated, but in the specific example 2 described next, an intent, an entity name, and a conversational sentence are used by using one document that also serves as the first document and the second document. One or more conversational sentence information that is a template of is generated.

(Specific example 2)
In this example, the document storage unit 101 stores, for example, a document file as shown in FIG. This document file is a kind of one document that doubles as a first document and a second document. This document file has a character string "expense settlement report" including an action name "expense settlement", and also has a character string "expense" which is a part of the action name. In addition, this document file also has the attribute name "employee". The action name "expense settlement" is one of one or more pre-registered intents, and the attribute name "employee" is one of one or more pre-registered entity names. ..

In the storage unit 10, for example, an intent name dictionary as shown in FIG. 18 is stored in advance. The intent name dictionary is a dictionary related to intent names. In the intent name dictionary, for example, one set or two or more sets of intent names and one or more vocabulary sets are registered. The registered intent name may be an action name. Therefore, the intent name dictionary may be called, for example, an action name dictionary.

The pair of intent name and one or more vocabulary is, for example, the pair of intent name "expense settlement" and one or more vocabulary "expense settlement, expense statement", intent name "daily work report" and one or more vocabulary " Daily work report, daily report, intent name "monthly work report" and one or more vocabulary "monthly work report, monthly report", intent name "working hours" and one or more vocabulary "working hours, working hours" , The intent name "overtime hours" and one or more vocabulary "overtime hours, overtime", the intent name "vacation days" and one or more vocabulary "vacation days, holidays", etc.

Further, in the storage unit 10, for example, an entity name dictionary as shown in FIG. 19 is also stored in advance. The entity name dictionary is a dictionary related to entity names. The registered entity name may be an attribute name. Therefore, the entity name dictionary may be called, for example, an attribute name dictionary. In the entity name dictionary, for example, one or more pairs of entity names and one or more vocabularies are stored.

A pair of an entity name and one or more vocabularies is, for example, a pair of an entity name "employee" and one or more vocabulary "employees", a pair of an entity name "workplace" and one or more vocabularies "workplace, work place". , The entity name "transportation cost" and one or more vocabulary "transportation cost", the entity name "working time" and one or more vocabulary "working time, working time", and the like.

First, the processing unit 30 extracts one or more character strings from the document file stored in the document storage unit 101. The character string is usually an array of two or more characters, but includes the case of only one character. The character string may also include symbols.

In this example, for example, the first character string "・ Be sure to get the confirmation mark of your boss on the approval list.", The second character string "・ Please be sure to fill in the settlement on this form on a monthly basis.", 3rd character string "[Expense settlement report]", 4th character string "Target period", 5th character string "June 1, 2016-June 30, 2016", 6th character string "Creator", 7th character string "Ichiro Nakamura", 8th character string "Employee CD", 9th character string "053", 10th character string "creation date", 11th character string "July 9, 2016", No. Twelve strings "re-examiner", thirteenth string "approver", fourteenth string "day", fifteenth string "apply", sixteenth string "expense", seventeenth The character string "temporary payment", the eighteenth character string "item", the nineteenth character string "amount", etc. are extracted.

Next, the processing unit 30 performs morphological analysis on each of the extracted one or more character strings, acquires one or two or more words or symbols constituting the character string, and has a part of speech of each of the one or more words. Or specify the type of symbol. Then, the processing unit 30 uses information such as the specified part of speech to exclude one or more character strings that satisfy a predetermined condition from the extracted one or more character strings.

The predetermined condition is, for example, the first condition that "the extracted character string is a sentence". Whether or not the first condition is satisfied is determined based on, for example, the type and number of words constituting the character string, the part of speech of the word at the end of the character string, or the type of the symbol at the end. The technique for determining whether or not a character string is a sentence is known, and detailed description thereof will be omitted.

Alternatively, the predetermined condition may be, for example, a second condition that "the extracted character string is a numerical value or a date". Whether or not the second condition is satisfied is determined based on, for example, the part of speech and the format of the character string. A technique for determining whether or not a character string is a numerical value or a date is known, and detailed description thereof will be omitted.

In this example, the processing unit 30 excludes the character strings that meet either the first or second condition from the extracted one or more character strings. Therefore, among the 19 character strings of the first to nineteenth, the five character strings of the first, second, fifth, ninth and eleventh are excluded.

Next, the processing unit 30 excludes the character strings corresponding to either the first or second condition from the extracted one or more character strings, and for each of the remaining one or more character strings, the character concerned. Determine if the column contains one or more of the symbols or punctuation marks. Then, when it is determined that one or more pieces of information among the symbols or punctuation marks are included, the processing unit 30 deletes the one or more pieces of information from the character string.

In this example, of the remaining 14 character strings of the third, fourth, sixth to eighth, tenth, twelfth to nineteenth, only the third character string has two symbols "[" and It contains "]" and the two symbols "[" and "]" are deleted from the third character string.

Next, the intent generation unit 301 determines the degree of similarity between the character string and each of the one or more intent names registered in the intent name dictionary, or the said, for each of the remaining one or more character strings. The similarity between the intent name and one or more vocabularies paired with each other is acquired, and the intent name corresponding to the highest similarity among the acquired one or more similarity is specified. The similarity may be, for example, the ratio of matching characters in the character string (for example, when y characters in x characters match, the similarity degree is “y / x”, etc.), or the cosine similarity degree or the like. , The type does not matter.

Next, the intent generation unit 301 determines whether or not the highest similarity is equal to or greater than or greater than the threshold value, and when it is determined that the highest similarity is equal to or greater than the threshold value or is greater than the threshold value, the intent generation unit 301 has the character string. Get the tent name.

If it is determined that the highest similarity is less than or equal to the threshold value, the intent generation unit 301 normally does not generate an intent. However, even when it is determined that the highest similarity is less than or equal to the threshold value, the intent generation unit 301 may generate an intent having the character string. The output unit 40 outputs the intent thus generated, and the reception unit 20 receives a user's instruction indicating whether or not the intent can be accumulated, and when the instruction indicating whether or not the intent can be accumulated is received, the storage unit 306 May accumulate the intent.

In the case of this example, the similarity is the ratio of matching characters between the character string and the entity name or vocabulary, and the threshold value is "1/2". The ratio of matching characters may be, for example, a value obtained by dividing the number of matching characters by the number of characters in a character string or the number of characters such as an entity name, whichever is larger. Alternatively, the similarity may be, for example, the number of matching characters between the character string and the entity name or vocabulary, and the acquisition method is not limited.

The intent generation unit 301 first acquires the similarity between the first character string "expense settlement report" and the six intent names registered in the intent name dictionary among the remaining 14 character strings. To do. Since 4 out of 5 characters of the first character string "expense settlement report" match the first intent name "accounting settlement", the similarity "4/5" is acquired.

Alternatively, the intent generation unit 301 includes each of the two vocabularies "accounting settlement" and "accounting statement" paired with the first character string "expense settlement report" and the first intent name "accounting settlement". You may get the similarity of. In this case, the degree of similarity "4/5" with the former vocabulary "accounting settlement" and the degree of similarity "2/5" with the latter vocabulary "accounting statement" are acquired. The first character string "Expense Report" does not match any of the other five intent names, nor does it match one or more vocabularies paired with each intent name. The similarity "0" is acquired for each of the five intent names and the like.

In the intent generation unit 301, the highest similarity "4/5" among the 6 similarity or 12 similarity acquired for the first character string "expense settlement report" is the threshold value "1/2". Since it is larger than, the intent name "expense settlement" corresponding to the highest similarity is specified, and the intent including the first character string "expense settlement report" is generated. The intent including the "expense settlement report" is, for example, the "expense settlement report", but may be the "expense settlement report intent", the "expense settlement report search intent", etc., regardless of the format. ..

The same processing as above is performed for each of the second and subsequent character strings, and the similarity between the 11th character string "expense" and the first intent name "accounting settlement" is "2/4". Since there is a threshold of "1/2" or more, an intent including the eleventh character string "expense" is further generated. The intent including "expense" is, for example, "expense", but may be "expense intent", "expense search intent", or the like. For other character strings, the intent is not generated because the similarity with the registered intent name or the like is "0".

The storage unit 306 stores one or more intents generated by the intent generation unit 301 in the intent storage unit 12a. In this example, the two generated intents "Expense Report" and "Expense" are accumulated.

Next, the entity generation unit 302 uses the remaining one or more character strings for each one or more character strings excluding the one or more character strings that are the target of intent generation by the intent generation unit 301. The similarity between the character string and one or more entity names registered in the entity name dictionary, or the similarity with one or more vocabularies paired with the entity name is acquired. Then, the entity generation unit 302 determines whether or not the highest similarity among the acquired 1 or more similarity is equal to or greater than the threshold value or greater than the threshold value, and when it is determined that the degree is equal to or greater than the threshold value or greater than the threshold value. Generates the entity name corresponding to the character string.

If it is determined that the highest degree of similarity is less than or equal to the threshold value, the entity generation unit 302 normally does not generate the entity name. However, even if it is determined that the highest similarity is less than or equal to the threshold value, the entity generation unit 302 may generate an entity name including the character string. The output unit 40 outputs the entity name generated in this way, and the reception unit 20 receives a user's instruction indicating whether or not the entity name can be stored, and when the instruction indicating whether or not the storage is possible is received, the storage unit 306 May accumulate the entity name.

In this example, of the remaining 14 character strings, for each of the 12 character strings excluding the two character strings "expense settlement" and "expense" that were the targets of intent generation, the character string and the entity name. The degree of similarity with each one or more entity names "employee", "work place", "transportation expense", "working time", etc. registered in the dictionary is acquired.

Specifically, first, regarding the first character string "target period", the entity generation unit 302 matches the fourth entity name "working hours" with one of the four characters, so that the similarity is "1 /". Get 4 ". The similarity "0" is acquired for each of the other entity names and the like. Next, the entity generation unit 302 determines whether or not the highest similarity “1/4” among the acquired 4 or 6 similarity is equal to or greater than the threshold value “1/2”. Here, since it is determined that the highest similarity "1/4" is less than the threshold value "1/2", the entity name including the first character string "target period" is not generated.

For each of the second and third character strings, no character matches any of the registered entity names, so similarity other than "0" is not acquired, and the entity name including the character string is not generated.

The fourth character string "employee CD" includes the fourth character string because the similarity with the first entity name "employee" is "2/4" and the threshold value is "1/2" or more. The entity name is retrieved. The entity name including the fourth character string is, for example, "employee CD", but may be "employee CD entity" or the like, and its format does not matter.

For each character string after the 5th, the similarity other than "0" is not acquired, and the entity name including the character string is not generated.

The storage unit 306 stores one or more entity names generated by the entity generation unit 302 in the entity storage unit 15a. In this example, the generated entity name "Employee entity is accumulated.

In the template storage unit 102, "show {} of the initial template" {} "that has two variables" {} "is stored, and the processing unit 30 stores the generated intent (for example," $ 1 "". ) Is associated with the second variable "{}" in "Show {} of the initial template" {} ", and the generated entity name (for example," $ 2 ") is associated with the first variable" {{. } ”, The template“ {$ 1} of {$ 2}, which has the variable “{$ 2}” with the entity name and the variable “$ 1” with the intent, is displayed. Generate.

In this example, the processing unit 30 associates the generated intent "expense settlement report" with the second variable in the initial template, and assigns the generated entity name "employee CD" to the first variable. By associating with the variable "{}" of, the template "Show {expense settlement statement} of {employee CD}" is generated. Further, the processing unit 30 associates the generated intent "expense" with the second variable in the initial template, and associates the generated entity name "employee CD" with the first variable "{}. By associating with "", the template "Show {expense} of {employee CD}" is generated.

The storage unit 306 displays the two generated templates "Show {expense settlement} of {employee CD}" and "Show {expense} of {employee CD}" with the specified intent name "expense". It is stored in the conversation text information storage unit 14a in association with "payment".

In this way, using the document file, the intent "Expense Report", the entity name "Employee CD", and the template "Show {Expense Report} of {Employee CD}" are generated, and the intent "Expense" is generated. , Entity name "employee CD", and template "show {expense reimbursement} of {employee CD}" are also generated.

In this example, for example, the entity mapping information storage unit 16 stores entity mapping information 5 (for example, “MST_SHAIN, SHAIN_CODE, employee CD entity”, etc.), and the entity generation unit 302 stores the entity mapping. Using the information 5, the attribute values "1", "2", etc. corresponding to the attribute name "SHAIN_CODE" are acquired from the table information 1 (that is, the employee master: see FIG. 4) identified by the table identifier "MST_SHAIN". Then, it may be stored in the entity storage unit 15a in association with the entity name "employee CD entity".

As a result, one or more entities "1", "2", etc. corresponding to the entity name "employee CD" are also generated by using the table information 1. The document file of FIG. 17 also has an attribute value "053" corresponding to the attribute name "employee CD", and the entity generation unit 302 uses this document file to correspond to the entity name "employee CD". You may generate the entity "053".

Further, the conversation sentence information generation unit 303 inserts each of the generated one or more entities into the variable {employee CD} of each of the two generated templates, thereby causing one or more conversation sentence information “employee code”. You may generate "Show me the expense report of 1", "Show me the expense report of employee code 2", "Show me the expense of employee code 1", "Show me the expense of employee code 2", etc. The storage unit 306 stores one or more generated conversational sentence information in the conversational sentence information storage unit 14a in association with the specified intent name “expense settlement”. Thereby, the intent is supported. One or more conversational sentence information is also generated.

As described above, according to the present embodiment, the document storage unit 101 is a kind of first document having an action name which is information managed for each business process and is information for specifying the business process, and is a design related to API. An API design document that is a document is stored, and one or more including a table that is a kind of a second document having an attribute name and one or more words corresponding to the attribute name, and a table identifier that identifies the table in the table storage unit 11a. Table information is stored, and a template that is a conversational sentence template and has one or more variables is stored in the template storage unit 102, and the information generator 100 acquires an action name from the API design document and acquires it. An intent that is information having an action name is generated, and for each one or more table information, one or more words corresponding to the attribute name and the attribute name are acquired from the table possessed by the table information, and the acquired attribute name. Generate one or more entities that are the entity name and one or more words acquired, acquire one entity from one or more entities generated for each one or more table information, and acquire one or more entities. By inserting into one or more variables of the template, one or more conversation sentence information which is the information of the conversation sentence is generated, the generated intents are accumulated, and one or more for each one or more generated entity names. It is information for converting words into API parameters from API design document and table information by accumulating by associating entities and accumulating one or more generated conversational sentence information by associating with intents. Intent, one or more entity information, and one or more conversational sentence information can be automatically generated.

Further, one or more intents are stored in the intent storage unit 12a, and the information generation device 100 extracts one or more API information which is information having an action name and is information about API from the API design document. By acquiring action names from one or more API information, generating intents having one or more acquired action names, and accumulating the generated one or more intents in the intent storage unit 12a, APIs are used. Intents can be automatically generated from the design document.

Further, one or more entities are stored in the entity storage unit 15a, and the information generation device 100 acquires an attribute name from the table information for each one or more table information and generates an entity name which is an attribute name. For each one or more generated entity names, obtain one or more words from the table corresponding to the entity name, generate one or more entities that are one or more words, and associate them with each one or more entity names. By accumulating one or more entities, which are one or more words obtained from the table corresponding to the entity name, in the entity storage unit 15a, the entity name and the one corresponding to the entity name are stored from each of the one or more tables. The above entities can be automatically generated.

Further, the conversation text information storage unit 14a stores one or more conversation text information, and the template storage unit 102 is a template having one or more variables associated with the entity names, and the positions or numbers of the variables are set. Two or more different templates are stored, and the information generator 100 stores one entity corresponding to the entity name associated with each one or more variables of the template for each of the two or more templates in the entity storage unit 15a. One or more conversation text information is generated by inserting the acquired entity at the position of one or more variables of the template for each of two or more templates obtained from, and one or more conversations generated. It is a template having one or more variables to which the entity name is associated by accumulating the sentence information in the conversation sentence information storage unit 14a in association with the intent, and two or more templates having different positions or numbers of variables. Various conversational sentence information can be automatically generated by using.

Further, one or more entity mapping information is stored in the entity mapping information storage unit 16a, and the information generation device 100 stores the table identifier and the table identified by the table identifier from the table information for each one or more table information. The translation item name, which is the attribute name in the above and is the information that identifies the entity to be converted, and one entity corresponding to the translation item are acquired, and the entity name corresponding to one entity is obtained from the entity storage unit 15a. By acquiring, generating entity mapping information having a table identifier, a translation item name, and an entity name, and accumulating one or more generated entity mapping information in the entity mapping information storage unit 16a, from one or more table information. , One or more entity mapping information having a table identifier, a translation item name and an entity name can be automatically generated.

Further, one or more PK items are stored in the PK item storage unit 17a, and the information generation device 100 stores a table identifier for identifying the table and a table primary key from the table information for each one or more table information. By acquiring the specified primary key identifier, generating PK items that are information having the table identifier and the primary key identifier, and accumulating the generated one or more PK items in the PK item storage unit 17a, one or more. One or more PK items having a table identifier and a primary key identifier can be automatically generated from the table information of.

The software that realizes the information processing device in this embodiment is, for example, the following program. That is, this program is a document storage unit that stores API design documents that are information managed for each business process, design documents having action names that are information that identifies business processes, and API design documents. 101, a table having an attribute name and one or more words corresponding to the attribute name, a table storage unit 11a for storing one or more table information including a table identifier that identifies the table, and a conversation sentence template. A computer that can access the template storage unit 102 that stores a template having one or more variables acquires an action name from the API design document and generates an intent that is information having the acquired action name. For each of the intent generation unit 301 and one or more table information, an attribute name and one or more words corresponding to the attribute name are acquired from the table of the table information, and the entity name which is the acquired attribute name, And, one entity is acquired from the entity generation unit 302 that generates one or more entities that are one or more acquired words, and the one or more entities that the entity generation unit 302 generates for each of the one or more table information. Then, by inserting the acquired one or more entities into one or more variables possessed by the template, the conversation sentence information generation unit 303 that generates one or more conversation sentence information that is the information of the conversation sentence, and the intent generation The intent generated by the unit 301 is accumulated, and one or more entities are associated with each one or more entity names generated by the entity generation unit 302 and accumulated, and the conversation sentence information generation unit 303 generates 1 This is a program for functioning as a storage unit 306 that stores the above conversational text information in association with the intent.

(Embodiment 3)
FIG. 16 is a block diagram of the information generator 200 according to the present embodiment. The information generation device 200 in the present embodiment replaces the intent generation unit 301 with the intent generation unit 301a in the information generation device 100 of the second embodiment. The intent generation unit 301a includes a table name extraction unit 3011a, an action name generation unit 3012a, and an intent generation unit 3013a.

The DB design document is stored in the document storage unit 101. The DB design document is a design document related to the database. The DB design document has, for example, the table name of one or more tables stored in the table storage unit 11a. Since the table name has been described in the second embodiment, it will not be repeated. In addition, the DB design document usually includes one or more attribute names included in the stored table. However, the structure of the DB design document does not matter.

For example, the intent generation unit 301a acquires a table name or attribute name from the DB design document stored in the document storage unit 101, and constitutes an action name having at least a part of the acquired table name or attribute name. , Generates an intent that is information with the configured action name.

Alternatively, the intent generation unit 301a acquires one or two or more attribute names from the DB design document, and configures an action name having at least a part of the attribute names for each of the acquired one or more attribute names. You may also generate an intent that is information with the configured action name.

The table name extracting means 3011a extracts one or more table names or one or more attribute names from the DB design document.

The action name generation means 3012a extracts at least a part from the table name or the attribute name for each of the extracted one or more table names or one or more attribute names, and generates an action name having the extracted part. To do.

The intent generation means 3013a generates an intent having one or more generated action names.

For example, the DB design document includes the table name "working time table", and the table name extracting means 3011a acquires the table name "working time table" and the like from the DB design document. The action name generation means 3012a extracts a part of "working hours" from the extracted table name "working hours table", and generates an action name "working hours inquiry" having the extracted part. The intent generation means 3013a generates an intent "working data inquiry intent" having the generated action name.

Alternatively, the DB design document has, for example, one or more attribute names "ZIKANGAI_TIME", "TOTAL_TIME", etc. included in the table (that is, working hours table) included in the table information 3 stored in the table storage unit 11a. The table name extracting means 3011a acquires the one or more attribute names "ZIKANGAI_TIME", "TOTAL_TIME", etc. from the DB design document.

Then, for the acquired attribute name "ZIKANGAI_TIME", the action name generating means 3012a constitutes an action name "overtime hours inquiry" including the "overtime hours" obtained by converting the attribute name, and the intent generating means 3013a is Generate an intent "Overtime Inquiry Intent" with the configured action name.

Further, for the acquired attribute name "TOTAL_TIME", the action name generating means 3012a constitutes an action name "working time data inquiry" including the "working hours" obtained by converting the attribute name, and the intent generating means 3013a , You may generate an intent "working time data inquiry intent" with the configured action name.

The operation of the other elements is the same as that of the second embodiment, and the description thereof will be omitted.

The operation of the information generator 200 is such that step S1501 is replaced with the following step S1501a (not shown) in the flowchart of FIG.

(Step S1501a) The intent generation unit 301a acquires the table name or attribute name from the DB design document stored in the document storage unit 101, and constitutes an action name having at least a part of the table name or attribute name. , Generates an intent that is information with the configured action name.

As described above, according to the present embodiment, the document storage unit 101 is a table name including information managed for each business process and which can form at least a part of an action name which is information for specifying the business process. Alternatively, a DB design document which is a kind of first document having an attribute name and is a design document related to a database is stored, and a second document having an attribute name and one or more words corresponding to the attribute name in the table storage unit 11a. A table that is a kind, and one or more table information including a table identifier that identifies the table are stored, and a template that is a conversational sentence template and has one or more variables is stored in the template storage unit 102, and is an information generator. Reference numeral 200 denotes a table name or an attribute name obtained from a DB design document, an action name having at least a part of the table name or the attribute name, and an intent which is information having the configured action name. For each of the above table information, one or more words corresponding to the attribute name and the attribute name are acquired from the table of the table information, the entity name which is the acquired attribute name, and the acquired one or more words 1 Conversation by generating the above entities, acquiring one entity from one or more entities generated for each one or more table information, and inserting the acquired one or more entities into one or more variables that the template has. One or more conversational sentence information that is sentence information is generated, the generated intents are accumulated, and one or more entities are associated with each one or more generated entity names and accumulated, and one or more generated entities. Intent, which is information for converting words into API parameters from DB design document and table information by accumulating conversation sentence information in association with intent, one or more entity information, and one or more conversation. Statement information can be automatically generated.

Further, one or more intents are stored in the intent storage unit 12a, and the information generation device 200 extracts one or more table names or one or more attribute names from the DB design document, and one or more table names or one or more table names or For each one or more attribute names, at least a part is extracted from the table name or the attribute name, an action name having the extracted part is generated, and an intent having each one or more generated action names is generated. By accumulating one or more generated intents in the intent storage unit 12a, intents can be automatically generated from the DB design document.

Further, the information generator 200 also exhibits the other effects described in the second embodiment.

The software that realizes the information processing device in this embodiment is, for example, the following program. That is, this program is a kind of first document having a table name or attribute name including information that can form at least a part of an action name that is information managed for each business process and is information that identifies the business process. Identifyes a document storage unit 101 that stores a DB design document, which is a design document related to a database, a table that is a kind of a second document having an attribute name and one or more words corresponding to the attribute name, and the table. A computer that can access the table storage unit 11a that stores one or more table information including the table identifier to be used, and the template storage unit 102 that is a conversational sentence template and stores a template having one or more variables. An intent generator that acquires a table name or attribute name from the DB design document, configures an action name having at least a part of the table name or attribute name, and generates an intent that is information having the configured action name. For 301a and each of the above 1 or more table information, the attribute name and one or more words corresponding to the attribute name are acquired from the table of the table information, and the acquired attribute name, the entity name, and the acquired. One entity is acquired from the entity generation unit 302 that generates one or more entities that are one or more words, and the one or more entities that the entity generation unit 302 generates for each of the one or more table information. By inserting one or more entities into one or more variables of the template, the conversation sentence information generation unit 303 and the intent generation unit 301a that generate one or more conversation sentence information which is the information of the conversation sentence are generated. One or more conversational sentences generated by the conversational sentence information generation unit 303, and one or more entities are associated with each one or more entity names generated by the entity generation unit 302. This is a program for functioning as a storage unit 306 that stores information in association with the intent.

(Embodiment 4)
The information processing device 300 in the present embodiment is obtained by adding the function of the information generation device 100 of the second embodiment to the information output device 500 of the first embodiment. The information processing device 300 may be considered as, for example, an information output device or an information generation device. Further, such a matter also applies to the information processing apparatus 400 of the fifth embodiment.

The information processing device 300 includes a storage unit 1, a reception unit 2, a processing unit 3, and an output unit 4. The storage unit 1 includes a table storage unit 11, an intent storage unit 12, an API information storage unit 13, a conversation text information storage unit 14, an entity storage unit 15, an entity mapping information storage unit 16, a PK item storage unit 17, and a day conversion information. It includes a storage unit 18, a document storage unit 101, and a template storage unit 102. The reception unit 2 includes a conversation text reception unit 21. The conversation sentence reception unit 21 includes a voice reception means 211 and a voice recognition means 212.

The processing unit 3 includes an intent determination unit 31, a conversation sentence information determination unit 32, an entity acquisition unit 33, a parameter acquisition unit 34, an API information acquisition unit 35, an inquiry information configuration unit 36, a search unit 37, and an intent generation unit 301. , Entity generation unit 302, conversational sentence information generation unit 303, entity mapping information generation unit 304, PK item generation unit 305, and storage unit 306. The parameter acquisition unit 34 includes a determination unit 341, a day information acquisition unit 342, an entity name acquisition unit 343, a translation item name acquisition unit 344, a table identifier acquisition unit 345, a primary key identifier acquisition unit 346, and a conversion parameter acquisition unit 347. .. The output unit 4 includes a search result output unit 41.

The intent generation unit 301 includes API information extraction means 3011, action name acquisition means 3012, and intent generation means 3013. The entity generation unit 302 includes an entity name generation means 3021 and an entity generation means 3022. The conversation sentence information generation unit 303 includes an entity acquisition unit 3031 and a conversation sentence generation unit 3032. The storage unit 306 includes an intent storage means 3061, an entity storage means 3062, a conversational sentence information storage means 3063, an entity mapping information storage means 3064, and a PK item storage means 3065.

Each of the above elements operates in the same manner as in the first and second embodiments. However, the storage destination of the storage unit 306 is the storage unit 1. Similarly, the storage destination of the intent storage means 3061 is the intent storage unit 12, the storage destination of the entity storage means 3062 is the entity storage unit 15, and the storage destination of the conversation text information storage means 3063 is the conversation text information storage unit. The storage destination of the entity mapping information storage means 3064 is the entity mapping information storage unit 16, and the storage destination of the PK item storage means 3065 is the PK item storage unit 17.

As described above, according to the present embodiment, the information processing apparatus 300 automatically generates intents, one or more entity information, and one or more conversational sentence information from the API design document and the table information, and obtains the same search result. Information can be searched by accepting conversational sentences that change dynamically even with inquiries.

(Embodiment 5)
The information processing device 400 according to the present embodiment is obtained by adding the function of the information generation device 200 of the third embodiment to the information output device 500 of the first embodiment. In the information processing device 300 of the fourth embodiment, the information processing device 400 replaces the intent generation unit 301 with the intent generation unit 301a.

As described above, according to the present embodiment, the information processing apparatus 400 automatically generates intent, one or more entity information, and one or more conversational sentence information from the DB design document and the table information, and obtains the same search result. Information can be searched by accepting conversational sentences that change dynamically even with inquiries.

As described above, the information generation device according to the present invention has an effect that information for converting words used in the organization into API parameters can be automatically generated by using one or more documents possessed in the organization. , Useful as an information generator, etc.

Claims

A first document stored in a document storage unit that stores one or more documents including an action name or a part of the action name which is information for specifying a business process, and a second document having an attribute name. An intent generation unit that generates an intent that is information that acquires an action name and identifies a business process using a document and is information that has the action name.
Using the second document, an entity generation unit that generates an entity name that is the attribute name, and
By associating the entity name generated by the entity generation unit with the variable of the initial template stored in the template storage unit in which the initial template of the conversation sentence having the variable is stored, the entity name has a corresponding variable. The processing unit that generates the template and
An information generation device including an intent generated by the intent generation unit, an entity name generated by the entity generation unit, and a storage unit for accumulating templates generated by the processing unit.
The second document further comprises one or more words corresponding to the attribute name.
The entity generation unit
Further generate one or more entities that are one or more words,
The accumulation part is
The information generation device according to claim 1, wherein one or more entities are further accumulated in association with the entity name.
By inserting each one or more entities out of one or more entities generated by the entity generation unit into the variables associated with the entity names in the template, one or more information of conversational sentences related to business processing. It is further equipped with a conversational sentence information generation unit that generates conversational sentence information.
The accumulation part is
The information generation device according to claim 2, wherein one or more conversational sentence information generated by the conversational sentence information generation unit is associated with the intent and further accumulated.
The initial template has one or more variables
The processing unit
One or more entity names generated by the entity generation unit are associated with one or more variables of the initial template.
The document storage unit is
A table having an attribute name and one or more words corresponding to the attribute name, and a table storage unit for storing one or more table information including a table identifier that identifies the table.
The first document is a design document having an action name, an API design document which is a design document related to API, or a design document having a table name or an attribute name including information which can form at least a part of the action name. Yes, it is a DB design document that is a design document related to the database,
The second document is information on each of one or more tables stored in the table storage unit.
The intent generator
The action name is acquired from the API design document, or the table name or attribute name is acquired from the DB design document, and the action name having at least a part of the table name or attribute name is configured and acquired or configured. Generate an intent that is information with the action name
The entity generation unit
For each of the above 1 or more table information, the attribute name and one or more words corresponding to the attribute name are acquired from the table possessed by the table information, the acquired entity name which is the attribute name, and the acquired one or more. Generate one or more entities that are words and
The conversation sentence information generation unit
The entity generation unit acquires one entity from one or more entities generated for each of the one or more table information, and the acquired one or more entities are associated with one or more entity names of the template. Generate one or more conversational text information by inserting it into one or more variables.
The accumulation part is
The intent generated by the intent generation unit is accumulated, and one or more entities are associated with each one or more entity names generated by the entity generation unit and accumulated, and the conversation sentence information generation unit generates the intent. The information generation device according to claim 3, wherein one or more conversational sentence information is stored in association with the intent.
Further provided with an intent storage unit for storing one or more intents,
The intent generator
An API information extraction means for extracting one or more API information which is information having an action name and is information about API from the API design document.
An action name acquisition means for acquiring an action name from each of the above 1 or more API information,
It is provided with an intent generation means for generating an intent having one or more action names acquired by the action name acquisition means.
The accumulation part is
The information generation device according to claim 4, further comprising an intent storage unit that stores one or more intents generated by the intent generation means in the intent storage unit.
Further provided with an intent storage unit for storing one or more intents,
The intent generator
A table name extraction means for extracting one or more table names or one or more attribute names from the DB design document, and
An action name generating means that extracts at least a part from the table name or the attribute name for each one or more table names or each one or more attribute names and generates an action name having the extracted part.
It is provided with an intent generating means for generating an intent having one or more action names generated by the action name generating means.
The accumulation part is
The information generation device according to claim 4, further comprising an intent storage unit that stores one or more intents generated by the intent generation means in the intent storage unit.
Further provided with an entity storage unit for storing one or more entities,
The entity generation unit
For each one or more table information, an entity name generation means that acquires an attribute name from the table information and generates an entity name that is the attribute name, and
For each one or more entity names generated by the entity name generation means, one or more words are acquired from the table corresponding to the entity name, and one or more entities that are the one or more words are generated. And equipped
The accumulation part is
A claim further comprising an entity storage means for accumulating one or more entities, which are one or more words acquired from a table corresponding to the entity name, in the entity storage unit in association with each of the one or more entity names. 5 or the information generator according to claim 6.
Further equipped with a conversational sentence information storage unit for storing one or more conversational sentence information,
In the template storage unit,
A template that has one or more variables associated with entity names, and stores two or more templates that differ in the position or number of variables.
The conversation sentence information generation unit
For each of the two or more templates, an entity acquisition means for acquiring one entity corresponding to the entity name associated with each one or more variables of the template from the entity storage unit.
For each of the two or more templates, a conversation sentence generation means that generates one or more conversation sentence information by inserting an entity acquired by the entity acquisition means at the position of one or more variables of the template. Equipped with
The accumulation part is
The information generation device according to claim 7, further comprising a conversational sentence information storage unit that associates one or more conversational sentence information generated by the conversational sentence generation means with the intent and stores it in the conversational sentence information storage unit.
An entity mapping information storage unit that stores one or more entity mapping information,
For each of the above 1 or more table information, from the table information, a table identifier, a translation item name which is an attribute name in the table identified by the table identifier and is information for specifying the entity to be converted, and the relevant translation item name. An entity mapping having the table identifier, the translation item name, and the entity name by acquiring one entity corresponding to the translation item and acquiring the entity name corresponding to the one entity from the entity storage unit. It also has an entity mapping information generator that generates information.
The accumulation part is
The information generation device according to claim 7 or 8, further comprising an entity mapping information storage unit that stores one or more entity mapping information generated by the entity mapping information generation unit in the entity mapping information storage unit.
A PK item storage unit that stores one or more PK items,
For each of the one or more table information, the table identifier that identifies the table and the primary key identifier that identifies the primary key of the table are acquired from the table information, and the table identifier and the primary key identifier are obtained. It further includes a PK item generation unit that generates PK items that are information to have.
The accumulation part is
The information generation device according to any one of claims 4 to 9, further comprising a PK item storage unit that stores one or more PK items generated by the PK item generation unit in the PK item storage unit.
A corpus production method realized by an intent generation unit, an entity generation unit, a processing unit, and a storage unit.
A document storage unit in which the intent generation unit stores one or more documents including an action name or a part of the action name which is information for specifying a business process, and a second document having an attribute name. Using the first document stored in, the intent generation step of acquiring the action name, specifying the business process, and generating the intent which is the information having the action name,
An entity generation step in which the entity generation unit generates an entity name, which is the attribute name, using the second document.
The processing unit associates the entity name generated by the entity generation unit with the variable of the initial template stored in the template storage unit in which the initial template of the conversation sentence having the variable is stored, so that the entity name corresponds. Processing steps to generate a template with variables with
A method for producing a corpus, wherein the storage unit includes an intent generated by the intent generation unit, an entity name generated by the entity generation unit, and a storage step for accumulating a template generated by the processing unit.
A computer that can access a document storage unit that stores one or more documents including an action name or a part of the action name that is information that identifies a business process, and a second document that has an attribute name.
Using the first document stored in the document storage unit, an intent generation unit that acquires an action name, is information for specifying a business process, and generates an intent that is information having the action name,
Using the second document, an entity generation unit that generates an entity name that is the attribute name, and
By associating the entity name generated by the entity generation unit with the variable of the initial template stored in the template storage unit in which the initial template of the conversation sentence having the variable is stored, the entity name has a corresponding variable. The processing unit that generates the template and
A program for functioning as a storage unit for accumulating intents generated by the intent generation unit, entity names generated by the entity generation unit, and templates generated by the processing unit.