WO2016127740A1 - 信息查询方法和设备 - Google Patents

信息查询方法和设备 Download PDF

Info

Publication number
WO2016127740A1
WO2016127740A1 PCT/CN2016/070332 CN2016070332W WO2016127740A1 WO 2016127740 A1 WO2016127740 A1 WO 2016127740A1 CN 2016070332 W CN2016070332 W CN 2016070332W WO 2016127740 A1 WO2016127740 A1 WO 2016127740A1
Authority
WO
WIPO (PCT)
Prior art keywords
entity
query
category
specified
database
Prior art date
Application number
PCT/CN2016/070332
Other languages
English (en)
French (fr)
Inventor
王杰雄
杨扬
富卫军
陈一宁
Original Assignee
广州神马移动信息科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广州神马移动信息科技有限公司 filed Critical 广州神马移动信息科技有限公司
Priority to US15/550,591 priority Critical patent/US10860632B2/en
Publication of WO2016127740A1 publication Critical patent/WO2016127740A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/242Query formulation
    • G06F16/243Natural language query formulation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33Querying
    • G06F16/3331Query processing
    • G06F16/334Query execution
    • G06F16/3344Query execution using natural language analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/951Indexing; Web crawling techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/242Query formulation
    • G06F16/2433Query languages
    • G06F16/2445Data retrieval commands; View definitions
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/242Query formulation
    • G06F16/2433Query languages
    • G06F16/2448Query languages for particular applications; for extensibility, e.g. user defined types
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/245Query processing
    • G06F16/2452Query translation
    • G06F16/24522Translation of natural language queries to structured queries
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/28Databases characterised by their database models, e.g. relational or object models
    • G06F16/284Relational databases
    • G06F16/285Clustering or classification
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33Querying
    • G06F16/332Query formulation
    • G06F16/3325Reformulation based on results of preceding query
    • G06F16/3326Reformulation based on results of preceding query using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages

Definitions

  • the present invention relates to the field of Internet, and in particular, to an information query method and device.
  • the user's search request will also be very accurate, and the query words entered will contain a large number of precise intents, and the answer needs to be returned directly at the time of the query. For example, entering “Andy Lau's height” needs to return “174CM”; input “stars over 180cm” need to return to the list of stars that meet this height condition; enter “Tang and Song eight people” need to return to “Liu Zongyuan” and others.
  • the traditional information query system mostly adopts a layered architecture of a storage layer, an intermediate query layer, and a natural language analysis layer.
  • the query language in the intermediate query layer of the prior art is too simple, and can not accurately analyze and dismember the user input request, so that the result obtained by the user query is complicated and not intuitive enough; or is too complicated, and makes the nature
  • the language parsing layer cannot parse complex query languages, is not friendly enough to the natural language parsing layer, and cannot obtain the query results required by the user.
  • the traditional information query system returns the query result by comparing the matching degree of the user's query words with the text of the included webpage, it does not conform to the user's query intention, and cannot accurately return the query result required by the user.
  • a technical problem to be solved by the present invention is to provide an information query method and device, which can convert an abstract query statement into a basic query statement or iterate a plurality of basic query statements, can cover most query intents, and can implement complex logic. And inferential queries, so that you can easily return the results of the query.
  • an information query method comprising:
  • a query operation of a plurality of basic query statements according to a basic query statement or an iteration is performed.
  • the user's natural language query text can be first analyzed intent, the text without querying is filtered out, and the query text conforming to the intent is converted into an abstract query language statement. Converting an abstract query to a basic query can override most of the query intent; or iterate over multiple basic queries.
  • Basic query statements can be designed as simpler queries, such as obtaining a more explicit (eg, single) output based on fewer (eg, only one or two) inputs, and the relationship between the input and output can be closer, or direct. Iteration is the activity of the repeated feedback process. The purpose is to implement the query target.
  • Each execution of a basic query statement is called an "iteration", and the result of each iteration is used as the input of the next iteration, so that complex implementation can be realized.
  • Logical and inferential queries Converts a complex abstract query statement into iterations of several simple basic query statements. Each iteration obtains a more direct output based on fewer inputs. From one iteration, it gradually reaches the final query target from the original input. Since the basic query statements are relatively simple, the requirements for the database (storage layer) are low and do not need to be stored in a complicated structure. In this way, it is possible to easily obtain a query result that is more in line with the user's intention and more accurate.
  • the basic query statement may include:
  • the entity reverse query statement is used to reversely query the corresponding entity according to the specified attribute filtering condition and the specified category;
  • the database comprises: an entity database, wherein the record for one entity in the entity database comprises an entity data field and a variable attribute field, wherein the entity data field stores entity data representing the entity, and the variable attribute field stores the description entity Entity attribute data of the attribute;
  • each record in the relational database includes two nodes and side information, wherein two entity data respectively representing two entities are stored in the two nodes, and two entities are stored in the side information The relationship between the relationships between the entities.
  • Entity data fields, variable attribute fields, and two nodes and side information can be indexed in the database to further improve the efficiency of the query.
  • the record for an entity in the entity database may further include a meta information field in which meta information related to the entity is stored, and the meta information is information that distinguishes the entity from other entities.
  • the entity data is determined based on the meta information.
  • the meta-information distinguishes different entities and entity data, especially different entities of the same entity name. Therefore, when the entity is queried, the related information of the entity can be accurately obtained, and the phenomenon that other information that does not belong to the query entity appears to cause the fuzzy query occurs.
  • the database may further include a category database in which a plurality of entity category data and category labels are correspondingly stored, the plurality of entity category data is divided into a plurality of levels, and the lower level entity category data is subordinate to Associated higher level entity category data.
  • a category database in which a plurality of entity category data and category labels are correspondingly stored, the plurality of entity category data is divided into a plurality of levels, and the lower level entity category data is subordinate to Associated higher level entity category data.
  • a category tag corresponding to the entity class data describing the category of the entity is stored.
  • entity data in the entity database that satisfies the following conditions is retrieved:
  • the category label is an entity category data indicating a specified category or a category label corresponding to an entity category data belonging to the specified category;
  • the corresponding entity attribute data satisfies the specified attribute filter condition.
  • the query operation of the inverse determination of the entity data can be performed by identifying the category tag and satisfying the filter condition.
  • an entity attribute defined for the entity category represented by the entity category data may be stored in association with each entity category data.
  • the steps to perform a query operation based on an entity reverse query statement include:
  • the query operation according to the entity reverse query statement is performed for the entity database.
  • an information query device comprising:
  • a first conversion device configured to convert a natural language query text input by a user into a structured abstract query language statement
  • a second conversion device configured to convert the abstract query language statement into a basic query statement or multiple basic query statements of the iteration
  • the querying device is configured to execute a query operation of the plurality of basic query statements according to the basic query statement or the iteration for the database prepared in advance.
  • the querying device comprises:
  • An entity information querying device configured to execute an entity information query statement to query information related to the specified entity
  • An entity attribute querying device configured to execute an entity attribute query statement to query a specified attribute of the specified entity
  • An entity reverse query device configured to execute an entity reverse query statement to reversely query a corresponding entity according to the specified attribute filtering condition and the specified category;
  • a related entity querying device configured to execute a related entity query statement to query an entity having a specified relationship with a specified entity
  • the inter-entity relationship query device is configured to execute an inter-entity relationship query statement to query a relationship between two specified entities.
  • the database comprises:
  • An entity database in which the records for an entity in the entity database include entity data fields and The variable attribute field and the meta information field store entity data representing the entity in the entity data field, the entity attribute data describing the attribute of the entity in the variable attribute field, and the entity information related in the meta information field Meta information, meta information is information that distinguishes an entity from other entities, and the query device determines entity data based on the meta information;
  • each record in the relational database includes two nodes and side information, wherein two entity data respectively representing two entities are stored in the two nodes, and two entities are stored in the side information.
  • the database further includes a category database in which a plurality of entity category data and category labels are correspondingly stored, and the plurality of entity category data is divided into a plurality of levels, and the lower level entity category data is subordinate to the associated higher Hierarchical entity category data.
  • a category tag corresponding to the entity class data describing the category of the entity is stored.
  • the entity reverse query device retrieves entity data in the entity database that satisfies the following conditions:
  • the category label is an entity category data indicating a specified category or a category label corresponding to an entity category data belonging to the specified category;
  • the corresponding entity attribute data satisfies the specified attribute filter condition.
  • an entity attribute defined for the entity category represented by the entity category data is stored in association with each entity category data.
  • the entity reverse query device performs a query operation according to the entity reverse query statement for the entity database in a case where the specified attribute related to the specified attribute filter condition belongs to the entity attribute defined for the specified category.
  • an information query method including:
  • the intent analysis of the natural language query text input by the user is converted into a structured query statement
  • the query operation is performed according to the basic query statement or the plurality of basic query statements of the iteration to determine the feature information of the entity corresponding to the natural language query text.
  • an embodiment of the present invention provides a first possible implementation of the third aspect.
  • a method in which the natural language query text input by the user is subjected to intent analysis, and converted into a structured query statement includes:
  • the embodiment of the present invention provides the second possible implementation manner of the third aspect, wherein, in the database prepared in advance, performing the query operation according to the basic query statement or the plurality of basic query statements of the iteration includes:
  • the corresponding entity is reversely queried according to the specified attribute filtering condition and the specified category;
  • the embodiment of the present invention provides a third possible implementation manner of the third aspect, wherein, in the database prepared in advance, performing the query operation according to the basic query statement or the plurality of basic query statements of the iteration includes:
  • the basic query field is used to query the entity data field, and/or the variable attribute field, in which the entity data representing the entity is stored, and in the variable attribute field, the entity attribute describing the attribute of the entity is stored.
  • each record includes two nodes and side information. Among them, two entity data respectively representing two entities are stored in the two nodes, and the side information is stored in the side information. An inter-entity relationship data that represents the relationship between two entities.
  • the embodiment of the present invention provides a fourth possible implementation manner of the third aspect, wherein, in the entity database, the entity data field is queried using the basic query statement, and/or the variable attribute field includes:
  • the meta-information field is queried using a basic query statement, and meta-information related to the entity is stored in the meta-information field, and the meta-information is information that distinguishes the entity from other entities.
  • the embodiment of the present invention provides a fifth possible implementation manner of the third aspect, wherein, in the database prepared in advance, performing the query operation according to the basic query statement or the plurality of basic query statements of the iteration further includes: :
  • the entity category data and the category label are queried using the basic query statement, the plurality of entity category data is divided into multiple levels, and the lower level entity category data is subordinated to the higher level entity category data associated with the same;
  • the meta information field in the record for the entity is queried in the entity database, and the meta information field stores a category tag corresponding to the entity category data describing the category of the entity.
  • the embodiment of the present invention provides a sixth possible implementation manner of the third aspect, wherein, in the database prepared in advance, the reverse querying the corresponding entity according to the specified attribute filtering condition and the specified category includes:
  • the query category label is an entity category data indicating a specified category, or a category label corresponding to the entity category data belonging to the specified category; the corresponding entity attribute data satisfies the specified attribute filtering condition.
  • the embodiment of the present invention provides a seventh possible implementation manner of the third aspect, further including:
  • the execution step When the specified attribute involved in the specified attribute filtering condition belongs to the entity attribute defined for the specified category, the execution step reversely queries the corresponding entity according to the specified attribute filtering condition and the specified category in the database prepared in advance.
  • an information query apparatus including:
  • a first conversion module configured to perform intent analysis on the natural language query text input by the user, and convert the data into a structured query statement
  • a second conversion module converting the query statement into a basic query statement or multiple basic query statements of the iteration
  • the query module is configured to perform a query operation according to the basic query statement or the plurality of basic query statements of the iteration in the database prepared in advance to determine the feature information of the entity corresponding to the natural language query text.
  • the user's natural language query text can be first analyzed intent, the text without querying is filtered out, and the query text conforming to the intent is converted into an abstract query language statement.
  • An abstract query statement is converted to a basic query statement that covers most of the query intent; or iterates over multiple basic query statements.
  • Basic query statements can be designed as simpler queries, such as obtaining a more explicit (eg, single) output based on fewer (eg, only one or two) inputs, and the relationship between the input and output can be closer, or direct. Iteration is the activity of the repeated feedback process.
  • the purpose is to implement the query target.
  • Each execution of a basic query statement is called an "iteration", and the result of each iteration is used as the input of the next iteration, so that complex implementation can be realized.
  • Logical and inferential queries Converts a complex abstract query statement into iterations of several simple basic query statements. Each iteration obtains a more direct output based on fewer inputs. From one iteration, it gradually reaches the final query target from the original input. Since the basic query statements are relatively simple, the requirements for the database (storage layer) are low and do not need to be stored in a complicated structure. In this way, it is possible to easily obtain a query result that is more in line with the user's intention and more accurate.
  • FIG. 1 is a schematic flowchart of an information query method according to an embodiment of the present invention.
  • 2 is a database structure that can be used by the query method in accordance with the present invention.
  • 3 is an improved database structure that can be used by the query method in accordance with the present invention.
  • FIG. 4 is a schematic block diagram of an information inquiry device in accordance with one embodiment of the present invention.
  • Figure 5 is a schematic block diagram of an optional internal structure of the querying device of Figure 4.
  • Figure 6 is a basic flow diagram provided by a query method in accordance with the present invention.
  • FIG. 1 is a schematic flowchart of an information query method according to an embodiment of the present invention.
  • step S100 the natural language query text input by the user is converted into a structured abstract query language statement.
  • the natural language query text input by the user includes related content such as a character or event that is input in the search engine, which may be a simple word or a word, or a detailed descriptive sentence. .
  • step S100 specifically, the user's natural language query text may first be analyzed intent, the query words without query intent are filtered out, and the query words conforming to the intent are parsed, the query type is obtained, and the structure is formed.
  • the query string is converted into an abstract query language statement.
  • step S200 the abstract query language statement is converted into a basic query statement or a plurality of basic query statements of the iteration.
  • the abstract query statement obtained in step S100 can be converted into a basic query statement or a plurality of basic query statements, wherein the basic query statement can finally obtain the query target through multiple iterations, and can cover most of the queries.
  • the intent that is, the ability to query a specified attribute of the target at each iteration, can determine the query target more accurately by determining a plurality of different attributes of the target.
  • Iteration is the activity of the repeated feedback process, the purpose of which is to finally achieve relatively complex query goals through relatively simple query steps.
  • Each iteration of the process is called an "iteration", and the result of each iteration is used as input to the next iteration, enabling complex logic and reasoning analysis.
  • step S300 the query operation result of the plurality of basic query sentences according to the basic query sentence or the iteration is executed for the database prepared in advance.
  • step S200 the conversion of the basic query statement or the iteration of the plurality of basic query statements to the abstract query statement may overwrite the main query intent and enable complex logic and reasoning analysis.
  • the entity information query statement can be used to query information related to a specified entity, for example, all attribute information of a specified entity can be queried.
  • the entity information in the abstract query statement can be identified, and the related information of the entity is queried.
  • the entity information query statement can be recorded as EE (entity information), and the identified entity information is "Andy Lau", which can be expressed as EE (Andy Lau), which will return the query operation result of Andy Lau's related information.
  • An entity attribute query statement can be used to query a specified attribute of a specified entity.
  • the entity attribute query statement can be recorded as EAA (entity information, attribute name), and the identification of "Andy Lau's height” can be expressed as EAA (Andy Lau, height), which will return the results of the query operation of Andy Lau's height number.
  • the entity reverse query statement can be used to reversely query the corresponding entity according to the specified attribute filter condition and the specified category.
  • the filter condition and category of a specified attribute in the abstract query statement can be identified, and the entity that meets the condition and category is reversely queried.
  • the entity reverse query statement can be recorded as ATE (specified attribute filter condition, specified category), and the recognition is "singer star height more than 180cm", which can be expressed as ATE (height > 180cm, singer), will return to match The result of the query operation of the conditional singer list.
  • a related entity query statement can be used to query an entity that has a specified relationship with a specified entity.
  • an entity in an abstract query statement and some specified relationship can be identified, and the query conforms to another entity that has a specified relationship with the entity.
  • the relevant entity query statement can be recorded as ERE (entity name, specified relationship name), and the identification of "Andy Lau's wife” can be expressed as ERE (Andy Lau, wife), will return to Zhu Liqian (Andy Lau's wife) query Operation result.
  • An inter-entity relationship query statement can be used to query the relationship between two specified entities.
  • the inter-entity relationship query statement can be recorded as EER (entity name 1, entity name 2), and the identification of "What is the relationship between Andy Lau and Tony Leung,” can be expressed as EER (Andy Lau, Tony Leung), will return to Liu Dehua and Tony Leung The result of the query operation between the list of relationships.
  • the above basic query statements do not have a certain execution order, but are selected and executed by the keywords in the identified abstract query statements.
  • the above basic query languages can also be executed iteratively. For example, the identification of "the height of Andy Lau's wife” can be expressed as EAA (ERE (Andy Lau, wife), height), will return to Zhu Liqian's height query operation results.
  • FIG. 2 is an example of a database structure that can be used in a query method in accordance with the present invention.
  • the database structure can include at least two parts, an entity database (an entity library) and a relational database (a relational library).
  • the record for an entity in the entity database includes an entity data field and a variable attribute field in which entity data representing the entity is stored, and in the variable attribute field, entity attribute data describing the attribute of the entity is stored.
  • the entity database can be used to perform a query operation according to the entity information query statement and the entity attribute query statement.
  • Variable attribute fields can be indexed in the entity database. In this way, the query can be executed more efficiently.
  • Each record in the relational database includes two nodes and side information, wherein two entity data respectively representing two entities are stored in the two nodes, and a relationship between the two entities is stored in the side information.
  • Inter-entity relationship data In some embodiments, two nodes can be divided into an ingress node and an egress node, respectively storing entity A and entity B. At this time, the information stored in the side information is directional relationship data.
  • relational database can be used to perform a query operation according to an inter-entity relationship query statement and a related entity query statement. Nodes and side information can be indexed separately in a relational database. In this way, the query can be executed more efficiently.
  • the record for an entity in the entity database may further include a meta information field in which meta information related to the entity is stored, and the meta information is information that distinguishes the entity from other entities.
  • Meta-information also referred to as "metadata,” is data that describes data, that is, descriptive information about data and information resources.
  • the entity data may be determined based on the meta information.
  • the entity data may be first determined according to the meta information in the meta information field, and each meta information is associated with one entity data. This distinguishes between different entities and entity data through meta-information, so that when the entity is queried, the relevant information of the entity can be accurately obtained, and the phenomenon that other information not belonging to the query entity appears to cause fuzzy query is avoided. happened.
  • different entities of the same entity name for example, several different characters (entities) are called "Andy Lau" (entity data), which requires each of the characters to be distinguished by the difference in meta-information of each character. Individual character information (entity attribute data).
  • Figure 3 illustrates an improved database structure that can be used by the query method in accordance with the present invention.
  • the database may also include a category database (category library).
  • category database category library
  • a plurality of entity category data and category labels are correspondingly stored, and the plurality of entity category data is divided into multiple levels (first level category 1, first level category 2, ...; second level category 1, level 2) Category 2, ...; Level 3 Category 1, Level 3 Category 2, ...; hence).
  • the lower level entity category data is subordinate to the higher level entity category data associated with it.
  • a category tag corresponding to the entity class data describing the category of the entity is stored.
  • entity data that satisfies the following conditions can be retrieved in the entity database:
  • the category label is an entity category data indicating a specified category or a category label corresponding to an entity category data belonging to the specified category;
  • the corresponding entity attribute data satisfies the specified attribute filter condition.
  • the entity category data corresponding to the "singer" in the category database belongs to, for example, the secondary category, belonging to the first-level category “entertainment star”, and the third-level category “continental singer", “Hong Kong singer”, “Taiwanese singer” belongs to the second-class category "Singer”.
  • the category label tab-II of the secondary category "Singer” and the category label tab-III1 corresponding to the three categories of "Singapore", “Hong Kong singer” and “Taiwanese singer” belonging to “Singer” can be obtained from the category database.
  • the category labels stored in the meta information field can then be viewed in the entity database. Find the entity data of the category tags stored in the meta information as tab-II, tab-III1, tab-III2, and tab-III3.
  • an entity attribute defined for the entity category indicated by the entity category data may be stored in association with each entity category data. Only the attributes corresponding to the entity category data of the primary category 1 are shown in FIG. In fact, the corresponding attribute can be set with each (or more) entity category data (each category).
  • the steps of performing a query operation according to an entity reverse query statement may include:
  • the query operation according to the entity reverse query statement is performed for the entity database.
  • a judgment condition is set between the specified attribute filter condition and the entity category data, and the reverse query statement is executed for the entity database when the specified attribute related to the specified attribute filter condition belongs to the entity attribute defined for the specified category. Query operation. In this way, the execution of redundant and inappropriate query operations is avoided, making the query more targeted and accurate.
  • the specified attribute filter condition “height over 180cm” involves the specified attribute "height” belonging to the entity attribute defined for the specified category "singer", then the database performs entity reverse query.
  • the query operation of the statement If the specified attribute filter condition is changed from “height over 180cm” to "floor area over 100 square meters”, the specified attribute "land area” involved does not belong to the entity attribute defined for the specified category "singer”, then the database The query operation of the entity reverse query statement is not executed.
  • FIG. 4 is a schematic block diagram of an information inquiry device in accordance with one embodiment of the present invention.
  • the information query device includes:
  • a first conversion device 100 configured to convert a natural language query text input by a user into a structured abstract query language statement
  • a second conversion device 200 configured to convert an abstract query language statement into a basic query statement or a plurality of basic query statements of an iteration
  • the querying apparatus 300 is configured to execute a query operation of a plurality of basic query statements according to a basic query statement or an iteration for a database prepared in advance.
  • FIG. 5 is a schematic block diagram of an optional internal structure of the query device 300 shown in FIG.
  • the query device includes:
  • the entity information querying apparatus 310 is configured to execute an entity information query statement to query information related to the specified entity;
  • the entity attribute querying unit 320 is configured to execute an entity attribute query statement to query a specified attribute of the specified entity
  • the entity reverse query device 330 is configured to execute an entity reverse query statement to reversely query the corresponding entity according to the specified attribute filtering condition and the specified category;
  • the related entity querying device 340 is configured to execute a related entity query statement to query an entity having a specified relationship with the specified entity;
  • the inter-entity relationship query device 350 is configured to execute an inter-entity relationship query statement to query a relationship between two specified entities.
  • a variety of selective query operations are implemented by performing different querying languages on different query languages.
  • the database available for the query device according to the invention may comprise: an entity database and a relational database.
  • the record for an entity in the entity database may include an entity data field and a variable attribute field, a meta information field in which entity data representing the entity is stored, and an entity describing the attribute of the entity is stored in the variable attribute field.
  • the attribute data stores meta information related to the entity in the meta information field, the meta information is information that distinguishes the entity from other entities, and the querying device determines the entity data based on the meta information.
  • Each record in the relational database may include two nodes and side information, wherein two entity data respectively representing two entities are stored in the two nodes, and the side information is stored between the two entities.
  • Relational entity-to-entity relationship data
  • the database may further include a category database in which a plurality of entity category data and category labels are correspondingly stored, and the plurality of entity category data is divided into a plurality of levels, and the lower level entity category data is subordinate to the associated Higher level entity category data.
  • a category tag corresponding to the entity class data describing the category of the entity is stored.
  • the entity reverse query device retrieves entity data in the entity database that satisfies the following conditions:
  • the category label is an entity category data indicating a specified category or a category label corresponding to an entity category data belonging to the specified category;
  • the corresponding entity attribute data satisfies the specified attribute filter condition.
  • the database may include one or more of an entity database, a relational database, and a category database to perform several query operations in conjunction with the query device, which may be a query for entity information, an entity attribute query, an entity reverse query, and a related Entity query, inter-entity relationship query, etc.
  • entity database a relational database
  • category database to perform several query operations in conjunction with the query device, which may be a query for entity information, an entity attribute query, an entity reverse query, and a related Entity query, inter-entity relationship query, etc.
  • entity attributes defined for the entity category represented by the entity category data are stored in association with each entity category data.
  • the entity reverse query device performs a query operation according to the entity reverse query statement for the entity database in a case where the specified attribute related to the specified attribute filter condition belongs to the entity attribute defined for the specified category.
  • the judgment condition is set between the entity reverse query device and the category database, which avoids the execution of redundant inappropriate query operations, making the query more targeted and more accurate.
  • An embodiment of the present invention provides an information query method, as shown in FIG. 6, including the following steps:
  • S601 performing intent analysis on the natural language query text input by the user, and converting into a structured query statement
  • the natural language query provided by the user needs to perform intent analysis, and the purpose of the intent analysis is to determine the query result expected by the natural language query.
  • users use natural language to communicate, or when they send statements to the network by inputting natural language, they usually do not contain valid information like computer code, but use a large number of conjunctions in the language. Words that have no practical meaning, such as repetitive modifiers. These words have no value to the computer that needs to perform the retrieval step, and these words are likely to cause the computer to incorrectly identify the incorrect meaning. Therefore, before determining the keywords used for the retrieval, it is necessary to use the intention analysis method to remove the words with no value in the natural language query text provided by the user, or filter them out, thereby obtaining a structured query statement.
  • the query statement is similar to the query code of the computer. In general, the structured query contains only valid recognition words.
  • the feature information of the entity can be understood as the meta information in the preamble embodiment.
  • step S602 is executed to convert the obtained structured query statement filtered by the intent analysis to generate one or more basic query statements.
  • the more important process in this step is to perform a "word segmentation" process on a complete structured query. That is, an original coherent sentence is cut, so that each word (ie, the basic query statement) obtained by the cutting can be independently used as a recognition word. Then, based on these identification words and preset computer programming methods, you can write or convert the basic query statement.
  • step S601 the intent analysis of the natural language query text input by the user is converted into a structured query statement, which can be divided into the following sub-steps:
  • the true meaning of the natural language query text provided by the user that is, the query intent
  • the query intent is determined by way of intention analysis.
  • a structured query statement is obtained.
  • Step S601 and step S602 are explained below with a simple example.
  • the system obtains the natural language query text input by the user, and the text is specifically: “Singer with a height of more than 180 cm”.
  • the system performs intent analysis on the natural language query text, removes the "" word in the sentence, and obtains the structured query statement "the height is more than 180cm singer".
  • the system converts the statements obtained after the intent analysis, and obtains multiple recognition words. After entity annotation of these words, it is obtained: height, over, 180cm, singer. At this time, these four recognition words are already available. It was used directly by the query system. The system can start the search work according to the content of these recognized words.
  • the system generates a query statement, the query statement: ATE (singer), (attr weight, ">60kg”).
  • the query statement is provided to the retrieval system of the database in a preset coding form, and the corresponding retrieval result can be obtained.
  • the method according to the invention may also be embodied as a computer program product comprising a computer readable medium on which is stored a computer program for performing the functions described above in the method of the invention.
  • the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or combinations of both.
  • each block of the flowchart or block diagram can represent a module, a program segment, or a portion of code, and a module, a program segment, or a portion of code includes one or more executables for implementing the specified logical functions. instruction.
  • the functions noted in the blocks may also occur in a different order than those illustrated in the drawings. For example, two consecutive blocks may be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts can be implemented in a dedicated hardware-based system that performs the specified function or operation. Or it can be implemented by a combination of dedicated hardware and computer instructions.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种信息查询方法和设备,该方法包括:将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句;将所述抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句(S200);针对预先准备的数据库,执行根据所述基本查询语句或所述迭代的多个基本查询语句的查询操作。可以对用户的自然语言查询文本首先进行意图分析,过滤掉无需查询的文本,并对符合意图的查询文本转化成抽象查询语言语句;再将抽象查询语句转换为基本查询语句或者迭代多个基本查询语句;可以覆盖大多数的查询意图,还能够实现复杂的逻辑和推理查询,这样,可以方便地返回查询结果。

Description

信息查询方法和设备 技术领域
本发明涉及互联网领域,特别涉及信息查询方法和设备。
背景技术
目前,人们对信息查询的精准度要求越来越高,往往需要得到相对于所请求的查询词最准确的答案。
而在实际应用中,用户的搜索请求也会是很精确的,其输入的查询词中会包含大量的精确意图,需要在查询时直接返回答案。例如:输入“刘德华的身高”需要返回“174CM”;输入“身高超过180cm的明星”需要返回满足此身高条件的明星列表;输入“唐宋八大家”需要返回“柳宗元”等人。
但是,传统的信息查询系统多采用存储层、中间查询层和自然语言解析层的分层架构。而现有技术的中间查询层中的查询语言或者过于简单,而不能够对用户输入的请求做出精确的分析和肢解,使得用户查询得到的结果复杂且不够直观;或者过于复杂,而使得自然语言解析层对复杂的查询语言无法解析,对自然语言解析层不够友好,不能够得到用户所需要的查询结果。
所以,传统的信息查询系统在通过比对用户的查询词和收录网页的文本匹配程度来返回查询结果时,并不能符合用户的查询意图,而不能精确地返回用户所需要的查询结果。
因此,需要一种能够方便地返回查询结果的信息查询方法和设备。
发明内容
本发明所要解决的一个技术问题是提供了一种信息查询方法和设备,将抽象查询语句转换为基本查询语句或者迭代多个基本查询语句,可以覆盖大多数的查询意图,还能够实现复杂的逻辑和推理查询,这样,可以方便地返回查询结果。
根据本发明的第一个方面,提供了一种信息查询方法,包括:
将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句;
将抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句;
针对预先准备的数据库,执行根据基本查询语句或迭代的多个基本查询语句的查询操作。
由此,可以对用户的自然语言查询文本首先进行意图分析,过滤掉无需查询的文本,并对符合意图的查询文本转化成抽象查询语言语句。再将抽象查询语句转换为基本查询语句,可以覆盖大多数的查询意图;或者迭代多个基本查询语句。基本查询语句可以设计为较为简单的查询,例如根据较少的(例如仅一个或两个)输入获得较为明确的(例如单个)输出,输入和输出之间的关系距离可以较近,或者说较为直接。迭代则是重复反馈过程的活动,其目的是为了实现查询目标,每执行一个基本查询语句称为一次“迭代”,而每一次迭代得到的结果会作为下一次迭代的输入,从而可以实现复杂的逻辑和推理查询。将一个复杂的抽象查询语句转换为若干个简单的基本查询语句的迭代,每一次迭代都根据较少的输入获得较为直接的输出,通过一次次迭代,从原始输入逐步到达最终查询目标。由于基本查询语句都比较简单,所以对于数据库(存储层)的要求较低,不需要以复杂的结构来进行存储。这样,使得有可能方便地得到更符合用户意图和更加精准的查询结果。
优选地,基本查询语句可以包括:
实体信息查询语句,用于查询与指定实体相关的信息;
实体属性查询语句,用于查询指定实体的指定属性;
实体反向查询语句,用于根据指定属性过滤条件和指定类别来反向查询对应的实体;
相关实体查询语句,用于查询与指定实体具有指定关系的实体;以及
实体间关系查询语句,用于查询两个指定实体之间的关系。
这样,可以预先设定不同的基本查询语句类型,并根据所接收到的抽象查询语言文本,解析出与其对应的基本查询语句的类型,然后以此类型进行查询操作。不同的基本查询语言和满足了用户对不同查询文本的精准查询的需求。这些基本查询语句的输入和输出之间都有直接的关联,容易 从数据库或网页中实现其查询目的。
优选地,数据库包括:实体数据库,实体数据库中针对一个实体的记录包括实体数据字段和可变属性字段,在实体数据字段中存储有表示实体的实体数据,在可变属性字段中存储有描述实体的属性的实体属性数据;
以及关系数据库,关系数据库中的每条记录包括两个节点和边信息,其中,在两个节点中分别存储有分别表示两个实体的两个实体数据,在边信息中存储有表示两个实体之间的关系的实体间关系数据。
这样,通过按照基本查询语句的输入数据与输出数据之间的关系来设置数据库的存储结构,使得能够更加方便快捷地执行基本查询语句。可以在数据库中对实体数据字段、可变属性字段、以及两个节点、边信息建立索引,以进一步提高查询的效率。
优选地,实体数据库中针对一个实体的记录还可以包括元信息字段,在元信息字段中存储有与实体相关的元信息,元信息是使实体区别于其他实体的信息。
在执行查询操作的步骤中,基于元信息来确定实体数据。
这样,作为实体数据中的核心信息数据,元信息,就将不同的实体和实体数据进行了区分,特别是相同实体名称的不同实体。以便,在对实体查询的时候可以准确的获得实体的相关信息,避免了其他不属于查询实体的信息出现而造成模糊查询的现象的发生。
优选地,数据库还可以包括类别数据库,在类别数据库中,对应地存储有多个实体类别数据和类别标签,多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据。
在实体数据库中针对实体的记录中的元信息字段中,存储有与描述实体的类别的实体类别数据对应的类别标签。
在执行根据实体反向查询语句的查询操作时,在实体数据库中检索满足下述条件的实体数据:
类别标签为表示指定类别的实体类别数据或者从属于表示指定类别的实体类别数据所对应的类别标签;并且
相应的实体属性数据满足指定属性过滤条件。
这样,可以通过识别到类别标签,以及满足过滤条件而执行反向确定实体数据的查询操作。
优选地,在类别数据库中,可以与每个实体类别数据关联地存储有针对该实体类别数据所表示的实体类别定义的实体属性。
执行根据实体反向查询语句的查询操作的步骤包括:
在指定属性过滤条件所涉及的指定属性属于为指定类别定义的实体属性的情况下,针对实体数据库执行根据实体反向查询语句的查询操作。
这样,在过滤条件和实体类别数据之间设立了判断条件,避免了多余的不适当的查询操作的执行,使得查询更加有针对性,更准确。
根据本发明的第二个方面,提供了一种信息查询设备,包括:
第一转换装置,用于将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句;
第二转换装置,用于将抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句;
查询装置,用于针对预先准备的数据库,执行根据基本查询语句或迭代的多个基本查询语句的查询操作。
优选地,查询装置包括:
实体信息查询装置,用于执行实体信息查询语句,以查询与指定实体相关的信息;
实体属性查询装置,用于执行实体属性查询语句,以查询指定实体的指定属性;
实体反向查询装置,用于执行实体反向查询语句,以根据指定属性过滤条件和指定类别来反向查询对应的实体;
相关实体查询装置,用于执行相关实体查询语句,以查询与指定实体具有指定关系的实体;以及
实体间关系查询装置,用于执行实体间关系查询语句,以查询两个指定实体之间的关系。
优选地,数据库包括:
实体数据库,实体数据库中针对一个实体的记录包括实体数据字段和 可变属性字段、元信息字段,在实体数据字段中存储有表示实体的实体数据,在可变属性字段中存储有描述实体的属性的实体属性数据,在元信息字段中存储有与实体相关的元信息,元信息是使实体区别于其他实体的信息,查询装置基于元信息来确定实体数据;
关系数据库,关系数据库中的每条记录包括两个节点和边信息,其中,在两个节点中分别存储有分别表示两个实体的两个实体数据,在边信息中存储有表示两个实体之间的关系的实体间关系数据;以及
数据库还包括类别数据库,在类别数据库中,对应地存储有多个实体类别数据和类别标签,多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据。
在实体数据库中针对实体的记录中的元信息字段中,存储有与描述实体的类别的实体类别数据对应的类别标签。
其中,实体反向查询装置在实体数据库中检索满足下述条件的实体数据:
类别标签为表示指定类别的实体类别数据或者从属于表示指定类别的实体类别数据所对应的类别标签;并且
相应的实体属性数据满足指定属性过滤条件。
优选地,在类别数据库中,与每个实体类别数据关联地存储有针对该实体类别数据所表示的实体类别定义的实体属性。
实体反向查询装置在指定属性过滤条件所涉及的指定属性属于为指定类别定义的实体属性的情况下,针对实体数据库执行根据实体反向查询语句的查询操作。
根据本发明的第三个方面,提供了一种信息查询方法,包括:
将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句;
将查询语句转换为基本查询语句或迭代的多个基本查询语句;
在预先准备的数据库中,根据基本查询语句或迭代的多个基本查询语句,进行查询操作,以确定自然语言查询文本所对应的实体的特征信息。
结合第三方面,本发明实施例提供了第三方面的第一种可能的实施方 式,其中,将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句包括:
对自然语言查询文本进行意图分析,确定自然语言查询文本的查询意图;
去除自然语言查询文本中,与查询意图不相符的文字内容;
将去除后的自然语言查询文本转化为结构化的查询语句。
结合第三方面,本发明实施例提供了第三方面的第二种可能的实施方式,其中,在预先准备的数据库中,根据基本查询语句或迭代的多个基本查询语句,进行查询操作包括:
在预先准备的数据库中,查询指定实体的指定属性;
或,在预先准备的数据库中,查询与指定实体相关的信息;
或,在预先准备的数据库中,根据指定属性过滤条件和指定类别来反向查询对应的实体;
或,在预先准备的数据库中,查询与指定实体具有指定关系的实体;
或,在预先准备的数据库中,查询两个指定实体之间的关系。
结合第三方面,本发明实施例提供了第三方面的第三种可能的实施方式,其中,在预先准备的数据库中,根据基本查询语句或迭代的多个基本查询语句,进行查询操作包括:
在实体数据库中,使用基本查询语句查询实体数据字段,和/或可变属性字段,在实体数据字段中存储有表示实体的实体数据,在可变属性字段中存储有描述实体的属性的实体属性数据;以及
在关系数据库中,使用基本查询语句查询记录,每条记录包括两个节点和边信息,其中,在两个节点中分别存储有分别表示两个实体的两个实体数据,在边信息中存储有表示两个实体之间的关系的实体间关系数据。
结合第三方面,本发明实施例提供了第三方面的第四种可能的实施方式,其中,在实体数据库中,使用基本查询语句查询实体数据字段,和/或可变属性字段包括:
实体数据库中,使用基本查询语句查询元信息字段,在元信息字段中存储有与实体相关的元信息,元信息是使实体区别于其他实体的信息。
结合第三方面,本发明实施例提供了第三方面的第五种可能的实施方式,其中,在预先准备的数据库中,根据基本查询语句或迭代的多个基本查询语句,进行查询操作还包括:
在类别数据库中,使用基本查询语句查询实体类别数据和类别标签,多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据;
在实体数据库中查询针对实体的记录中的元信息字段,元信息字段中存储有与描述实体的类别的实体类别数据对应的类别标签。
结合第三方面,本发明实施例提供了第三方面的第六种可能的实施方式,其中,在预先准备的数据库中,根据指定属性过滤条件和指定类别来反向查询对应的实体包括:
查询类别标签为表示指定类别的实体类别数据,或者查询从属于表示指定类别的实体类别数据所对应的类别标签;相应的实体属性数据满足指定属性过滤条件。
结合第三方面,本发明实施例提供了第三方面的第七种可能的实施方式,其中,还包括:
当在指定属性过滤条件所涉及的指定属性属于为指定类别定义的实体属性时,执行步骤在预先准备的数据库中,根据指定属性过滤条件和指定类别来反向查询对应的实体。
根据本发明的第四个方面,提供了一种信息查询装置,包括:
第一转换模块,用于将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句;
第二转换模块,将查询语句转换为基本查询语句或迭代的多个基本查询语句;
查询模块,用于在预先准备的数据库中,根据基本查询语句或迭代的多个基本查询语句,进行查询操作,以确定自然语言查询文本所对应的实体的特征信息。
由此,可以对用户的自然语言查询文本首先进行意图分析,过滤掉无需查询的文本,并对符合意图的查询文本转化成抽象查询语言语句。再将 抽象查询语句转换为基本查询语句,可以覆盖大多数的查询意图;或者迭代多个基本查询语句。基本查询语句可以设计为较为简单的查询,例如根据较少的(例如仅一个或两个)输入获得较为明确的(例如单个)输出,输入和输出之间的关系距离可以较近,或者说较为直接。迭代则是重复反馈过程的活动,其目的是为了实现查询目标,每执行一个基本查询语句称为一次“迭代”,而每一次迭代得到的结果会作为下一次迭代的输入,从而可以实现复杂的逻辑和推理查询。将一个复杂的抽象查询语句转换为若干个简单的基本查询语句的迭代,每一次迭代都根据较少的输入获得较为直接的输出,通过一次次迭代,从原始输入逐步到达最终查询目标。由于基本查询语句都比较简单,所以对于数据库(存储层)的要求较低,不需要以复杂的结构来进行存储。这样,使得有可能方便地得到更符合用户意图和更加精准的查询结果。
附图说明
通过结合附图对本公开示例性实施方式进行更详细的描述,本公开的上述以及其它目的、特征和优势将变得更加明显,其中,在本公开示例性实施方式中,相同的参考标号通常代表相同部件。
图1是根据本发明的一个实施例的信息查询方法的示意性流程图。
图2是根据本发明的查询方法可以使用的数据库结构。
图3是根据本发明的查询方法可以使用的改进的数据库结构。
图4是根据本发明的一个实施例的信息查询设备的示意性方框图。
图5是图4中的查询装置的可选内部结构示意性方框图;
图6是根据本发明的查询方法所提供的基本流程图。
具体实施方式
下面将参照附图更详细地描述本公开的优选实施方式。虽然附图中显示了本公开的优选实施方式,然而应该理解,可以以各种形式实现本公开而不应被这里阐述的实施方式所限制。相反,提供这些实施方式是为了使本公开更加透彻和完整,并且能够将本公开的范围完整地传达给本领域的技术人员。
图1是根据本发明的一个实施例的信息查询方法的示意性流程图。
首先,在步骤S100,将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句。
用户输入的自然语言查询文本包括在搜索引擎中所输入的想要查询的人物或事件等的相关内容,这其中可以是简单的几个字或者词语,当然也可以是一句详细的描述性的句子。
在步骤S100中,具体说来,可以对用户的自然语言查询文本首先进行意图分析,过滤掉无查询意图的查询词,并对符合意图的查询词进行解析,得出其查询类型,并形成结构化查询串,从而转化成抽象查询语言语句。
然后,在步骤S200,将抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句。
在S200步骤中可以将在步骤S100中所得到的抽象查询语句转换为基本查询语句或者迭代多个基本查询语句,其中基本查询语句可以通过多次迭代而最终得到查询目标,可以覆盖大多数的查询意图,也就是每次迭代的时候能够查询到目标的一种指定属性,通过确定目标的多种不同属性,进而能够更加准确的确定查询目标。迭代是重复反馈过程的活动,其目的是为了通过相对多次简单的查询步骤最终实现相对复杂查询目标。每一次对过程的重复称为一次“迭代”,而每一次迭代得到的结果会作为下一次迭代的输入,从而可以实现复杂的逻辑和推理的分析。通过将一个复杂的抽象查询语句转换为若干个简单的基本查询语句的迭代,每一次迭代都根据较少的输入获得较为直接的输出,通过一次次迭代,从原始输入逐步到达最终查询目标。
由于基本查询语句都比较简单,所以对于数据库(存储层)的要求较低,不需要以复杂的结构来进行存储,也就是不再需要复杂的检索方式来完成检索任务。这样,使得有可能方便地得到更符合用户意图和更加精准的查询结果。
这样,在步骤S300,针对预先准备的数据库,执行根据基本查询语句或迭代的多个基本查询语句的查询操作结果。
通过如图1所示的方法,可以方便地得到更符合用户意图和更加精准的查询结果。
在步骤S200中,对抽象查询语句的转换基本查询语句或迭代多个基本查询语句,可以覆盖主要的查询意图,并且可以实现复杂的逻辑和推理的分析。
下面描述几种可以采用的基本查询语句。
实体信息查询语句可以用于查询与指定实体相关的信息,例如可以查询指定实体的所有属性信息。
这里,可以识别出抽象查询语句中的实体信息,并查询实体的相关信息。例如,实体信息查询语句可以记为EE(实体信息),而所识别到的实体信息是“刘德华”,则可以表示为EE(刘德华),将返回刘德华的相关信息的查询操作结果。
实体属性查询语句可以用于查询指定实体的指定属性。
例如,实体属性查询语句可以记为EAA(实体信息,属性名),而识别到的是“刘德华的身高”,可以表示为EAA(刘德华,身高),将返回刘德华的身高数字的查询操作结果。
实体反向查询语句可以用于根据指定属性过滤条件和指定类别来反向查询对应的实体。
这里,可以识别出抽象查询语句中的指定的某种属性的过滤条件和类别,而反向查询出符合该条件和类别的实体。例如:实体反向查询语句可以记为ATE(指定属性过滤条件,指定类别),而识别到的是“身高超过180cm的歌星”,可以表示为ATE(身高>180cm,歌星),将返回符合该条件的歌星列表的查询操作结果。
相关实体查询语句可以用于查询与指定实体具有指定关系的实体。
这里,可以识别出抽象查询语句中的实体和某种指定关系,而查询符合这种与该实体的指定关系的另一实体。例如:相关实体查询语句可以记为ERE(实体名,指定关系名),而识别到的是“刘德华的妻子”,可以表示为ERE(刘德华,妻子),将返回朱丽倩(刘德华的妻子)的查询操作结果。
实体间关系查询语句可以用于查询两个指定实体之间的关系。
这里,可以识别出抽象查询语句中的两个特定的实体,而查询符合这 种与该两个实体之间存在的关系。例如:实体间关系查询语句可以记为EER(实体名1,实体名2),而识别到的是“刘德华和梁朝伟有什么关系”,可以表示为EER(刘德华,梁朝伟),将返回刘德华和梁朝伟之间关系列表的查询操作结果。
另一方面,上述的几种基本查询语句并没有一定的执行顺序,而是通过所识别到的抽象查询语句中的关键词而选择执行,当然,上述几种基本查询语言还可以迭代执行。例如:所识别到“刘德华妻子的身高”,可以表示为EAA(ERE(刘德华,妻子),身高),将返回朱丽倩的身高查询操作结果。
当然,基本查询语句的子语句也并不仅限于上述所列举的几种,几种基本语句的迭代执行也不仅限于上述举例的迭代方式,满足其他的所需要的逻辑查询关系的基本查询语句及迭代执行,均可应用在其中。
图2是可用于根据本发明的查询方法的数据库结构示例。
数据库结构至少可以包括两部分,实体数据库(实体库)和关系数据库(关系库)。
实体数据库中针对一个实体的记录包括实体数据字段和可变属性字段,在实体数据字段中存储有表示实体的实体数据,在可变属性字段中存储有描述实体的属性的实体属性数据。
这里,实体数据库可以用于执行根据实体信息查询语句和实体属性查询语句的查询操作。可以在实体数据库中对可变属性字段建立索引。这样,可以更加高效地执行查询。
关系数据库中的每条记录包括两个节点和边信息,其中,在两个节点中分别存储有分别表示两个实体的两个实体数据,在边信息中存储有表示两个实体之间的关系的实体间关系数据。在一些实施例中,两个节点可以区分为入节点和出节点,分别存储实体A和实体B。此时边信息中存储的则是有方向性的关系数据。
这里,关系数据库可以用于执行根据实体间关系查询语句和相关实体查询语句的查询操作。可以在关系数据库中对节点和边信息分别建立索引。这样,可以更加高效地执行查询。
进一步地,实体数据库中针对一个实体的记录还可以包括元信息字段,在元信息字段中存储有与实体相关的元信息,元信息是使实体区别于其他实体的信息。元信息也可以称为“元数据”,是描述数据的数据,即对数据及信息资源的描述性信息。
在执行查询操作的步骤中,可以基于元信息来确定实体数据。
这样,针对基本查询语句,可以首先根据元信息字段中的元信息确定实体数据,每一个元信息都关联着一个实体数据。这就将不同的实体和实体数据之间通过元信息进行了区分,以便在对实体查询的时候可以准确的获得实体的相关信息,避免了其他不属于查询实体的信息出现而造成模糊查询的现象的发生。特别是,相同实体名称的不同实体,例如,几个不同的人物(实体)都叫“刘德华”(实体数据),这就需要通过每个人物的元信息的不同,将他们各自区分,以得到各自的人物信息(实体属性数据)。
图3示出了根据本发明的查询方法可以使用的改进的数据库结构。
如图3所示,除了实体数据库和关系数据库,数据库还可以包括类别数据库(类别库)。
在类别数据库中,对应地存储有多个实体类别数据和类别标签,多个实体类别数据被划分为多个层次(一级类别1、一级类别2、……;二级类别1、二级类别2、……;三级类别1、三级类别2、……;……)。较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据。
在实体数据库中针对实体的记录中的元信息字段中,存储有与描述实体的类别的实体类别数据对应的类别标签。
在执行根据实体反向查询语句的查询操作时,可以在实体数据库中检索满足下述条件的实体数据:
类别标签为表示指定类别的实体类别数据或者从属于表示指定类别的实体类别数据所对应的类别标签;并且
相应的实体属性数据满足指定属性过滤条件。
例如,假设要执行“身高超过180cm的歌星”的查询操作。
假设类别数据库中“歌星”对应的实体类别数据例如属于二级类别,从属于一级类别“娱乐明星”,并且三级类别“大陆歌星”、“香港歌星”、 “台湾歌星”从属于二级类别“歌星”。
可以从类别数据库获知二级类别“歌星”的类别标签tab-II以及从属于“歌星”的三级类别“大陆歌星”、“香港歌星”、“台湾歌星”各自对应的类别标签tab-III1、tab-III2、tab-III3。
然后,可以在实体数据库中查看在元信息字段中存储的类别标签。找出元信息中存储的类别标签为tab-II、tab-III1、tab-III2、tab-III3的实体数据。
然后,查看这些通过类别标签找到的实体数据对应的可变属性数据中关于身高的属性数据满足条件“超过180cm”的实体数据。
这样可以从数据库中反查出“身高超过180cm的歌星”。
另外,如图3所示,在类别数据库中,可以与每个实体类别数据关联地存储有针对该实体类别数据所表示的实体类别定义的实体属性。图3中只示出了与一级类别1的实体类别数据对应的属性。事实上,可以与每一个(或者多个)实体类别数据(每一个类别)设置对应的属性。
执行根据实体反向查询语句的查询操作的步骤可以包括:
在指定属性过滤条件所涉及的指定属性属于为指定类别定义的实体属性的情况下,针对实体数据库执行根据实体反向查询语句的查询操作。
这里,在指定属性过滤条件和实体类别数据之间设立了判断条件,是当指定属性过滤条件所涉及的指定属性属于为指定类别定义的实体属性的情况下,才针对实体数据库执行反向查询语句的查询操作。这样,避免了多余的不适当的查询操作的执行,使得查询更加有针对性,更准确。
例如:在“身高超过180cm的歌星”的这一例子中,指定属性过滤条件“身高超过180cm”涉及指定属性“身高”属于为指定类别“歌星”定义的实体属性,则数据库执行实体反向查询语句的查询操作。若将指定属性过滤条件从“身高超过180cm”改为“占地面积超过100平方米”,其中所涉及的指定属性“占地面积”不属于为指定类别“歌星”定义的实体属性,则数据库不执行实体反向查询语句的查询操作。
上面参考图1-3详细描述了信息查询方法。下面参照附图描述信息查询设备。
下面描述的设备很多功能分析与上面参考图1-3描述的相应方法步骤的功能相同。为了避免重复,这里重点描述设备具有的装置结构,而对一些细节则不再赘述,可以参考上文的相关描述。
图4是根据本发明的一个实施例的信息查询设备的示意性方框图。
如图4所示,信息查询设备,包括:
第一转换装置100,用于将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句;
第二转换装置200,用于将抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句;
查询装置300,用于针对预先准备的数据库,执行根据基本查询语句或迭代的多个基本查询语句的查询操作。
通过图4的所示设备,可以覆盖大多数的查询意图,实现复杂的逻辑和推理查询,方便地得到更符合用户意图和更加精准的查询结果。
图5是图4所示的查询装置300的可选内部结构示意性方框图。
如图5所示,查询装置包括:
实体信息查询装置310,用于执行实体信息查询语句,以查询与指定实体相关的信息;
实体属性查询装置320,用于执行实体属性查询语句,以查询指定实体的指定属性;
实体反向查询装置330,用于执行实体反向查询语句,以根据指定属性过滤条件和指定类别来反向查询对应的实体;
相关实体查询装置340,用于执行相关实体查询语句,以查询与指定实体具有指定关系的实体;以及
实体间关系查询装置350,用于执行实体间关系查询语句,以查询两个指定实体之间的关系。
通过对不同查询装置对不同的查询语言的执行,实现多种选择性的查询操作。
如上所述,可用于根据本发明的查询设备的数据库可以包括:实体数据库和关系数据库。
实体数据库中针对一个实体的记录可以包括实体数据字段和可变属性字段、元信息字段,在实体数据字段中存储有表示实体的实体数据,在可变属性字段中存储有描述实体的属性的实体属性数据,在元信息字段中存储有与实体相关的元信息,元信息是使实体区别于其他实体的信息,查询装置基于元信息来确定实体数据。
关系数据库中的每条记录可以包括两个节点和边信息,其中,在两个节点中分别存储有分别表示两个实体的两个实体数据,在边信息中存储有表示两个实体之间的关系的实体间关系数据。
该数据库还可以包括类别数据库,在类别数据库中,对应地存储有多个实体类别数据和类别标签,多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据。
在实体数据库中针对实体的记录中的元信息字段中,存储有与描述实体的类别的实体类别数据对应的类别标签。
其中,实体反向查询装置在实体数据库中检索满足下述条件的实体数据:
类别标签为表示指定类别的实体类别数据或者从属于表示指定类别的实体类别数据所对应的类别标签;并且
相应的实体属性数据满足指定属性过滤条件。
这里,数据库可以包括实体数据库,关系数据库和类别数据库中的一种或几种,以配合查询装置完成几种查询操作,可以是对实体信息的查询,实体属性的查询,实体反向查询,相关实体查询,实体间关系查询等。具体的查询步骤流程参见上文的对应位置处的详细描述。
在类别数据库中,与每个实体类别数据关联地存储有针对该实体类别数据所表示的实体类别定义的实体属性。
实体反向查询装置在指定属性过滤条件所涉及的指定属性属于为指定类别定义的实体属性的情况下,针对实体数据库执行根据实体反向查询语句的查询操作。
这里,在实体反向查询装置和类别数据库之间设立了判断条件,避免了多余的不适当的查询操作的执行,使得查询更加有针对性,更准确。具 体的操作步骤流程参见上文的对应位置处的详细描述。
本发明实施例一个实施例提供了一种信息查询方法,如图6所示,包括如下步骤:
S601,将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句;
S602,将查询语句转换为基本查询语句或迭代的多个基本查询语句;
S603,在预先准备的数据库中,根据基本查询语句或迭代的多个基本查询语句,进行查询操作,以确定自然语言查询文本所对应的实体的特征信息。
步骤S601,需要将用户所提供的自然语言查询本文进行意图分析,进行意图分析的目的是确定自然语言查询本文所期望得出的查询结果。用户在使用自然语言进行交流,或者是通过输入自然语言的方式在向网络端发送语句的时候,通常不会如同计算机代码一样,只包含有有效的信息,而是在语言中大量的使用连接词、重复化的修饰词等无实际意义的词语。这些词语对于需要执行检索步骤的计算机而言是没有任何价值的,并且,这些词语很可能会导致计算机错误的识别出不正确的含义。因此,在确定检索所使用的关键词之前,需要使用意图分析的方式,将用户所提供的自然语言查询文本中没有价值的词语去除,或者说是过滤掉,进而得到结构化的查询语句,此种查询语句已经类似于计算机的查询代码,一般情况下,结构化的查询语句中只包含有有效的识别词语。其中,实体的特征信息可以理解为前序实施例中的元信息。
之后,执行步骤S602,将使用意图分析过滤后的所得到的结构化查询语句进行转换,进而生成一个或者多个基本查询语句。该步骤中比较重要的过程是对一个完整的结构化的查询语句进行“分词”处理。也就是将一个原本连贯的语句进行切割,进而使切割所得出的每个词语(即基本查询语句)均能够独立的作为一个识别词语。之后,基于这些识别词语和预设好的计算机编程方式,便可以编写出,或者是转化出基本查询语句了。
具体而言,步骤S601,将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句,可以分为如下子步骤:
对自然语言查询文本进行意图分析,确定自然语言查询文本的查询意图;
去除自然语言查询文本中,与查询意图不相符的文字内容;
将去除后的自然语言查询文本转化为结构化的查询语句。
也就是,通过意图分析的方式来确定用户所提供的自然语言查询文本所要表达的真实含义,即查询意图。之后将该自然语言查询文本中与查询意图不相符的内容删除,便得到了结构化的查询语句。
下面以一个简单实例来说明步骤S601和步骤S602。
1,系统获取用户所输入的自然语言查询文本,该文本具体为:“身高超过180cm的歌星”。
2,系统对自然语言查询文本进行意图分析,去除该句话中的“的”字,得到结构化查询语句“身高超过180cm歌星”。
3,系统将意图分析后所得到的语句进行转化,得到了多个识别词语,在对这些词语进行实体标注后,得到:身高,超过,180cm,歌手,此时,这四个识别词语已经可以被查询系统直接使用了。系统便可以按照这些识别词语的内容开始检索工作。
4,系统生成查询语句,查询语句:ATE(歌手),(attr体重,“>60kg”)。将该查询语句以预设的编码形式提供给数据库的检索系统,便可以得到相应的检索结果。
至此,已经详细地描述了根据本发明的信息查询方法和设备。
此外,根据本发明的方法还可以实现为一种计算机程序产品,该计算机程序产品包括计算机可读介质,在该计算机可读介质上存储有用于执行本发明的方法中限定的上述功能的计算机程序。本领域技术人员还将明白的是,结合这里的公开所描述的各种示例性逻辑块、模块、电路和算法步骤可以被实现为电子硬件、计算机软件或两者的组合。
附图中的流程图和框图显示了根据本发明的多个实施例的系统和方法的可能实现的体系架构、功能和操作。在这点上,流程图或框图中的每个方框可以代表一个模块、程序段或代码的一部分,模块、程序段或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注 意,在有些作为替换的实现中,方框中所标记的功能也可以以不同于附图中所标记的顺序发生。例如,两个连续的方框实际上可以基本并行地执行,它们有时也可以按相反的顺序执行,这依所涉及的功能而定。也要注意的是,框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合,可以用执行规定的功能或操作的专用的基于硬件的系统来实现,或者可以用专用硬件与计算机指令的组合来实现。
以上已经描述了本发明的各实施例,上述说明是示例性的,并非穷尽性的,并且也不限于所披露的各实施例。在不偏离所说明的各实施例的范围和精神的情况下,对于本技术领域的普通技术人员来说许多修改和变更都是显而易见的。本文中所用术语的选择,旨在最好地解释各实施例的原理、实际应用或对市场中的技术的改进,或者使本技术领域的其它普通技术人员能理解本文披露的各实施例。

Claims (19)

  1. 一种信息查询方法,包括:
    将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句;
    将所述抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句;
    针对预先准备的数据库,执行根据所述基本查询语句或所述迭代的多个基本查询语句的查询操作。
  2. 根据权利要求1所述的方法,其中,所述基本查询语句包括:
    实体信息查询语句,用于查询与指定实体相关的信息;
    实体属性查询语句,用于查询指定实体的指定属性;
    实体反向查询语句,用于根据指定属性过滤条件和指定类别来反向查询对应的实体;
    相关实体查询语句,用于查询与指定实体具有指定关系的实体;以及
    实体间关系查询语句,用于查询两个指定实体之间的关系。
  3. 根据权利要求2所述的方法,其中,所述数据库包括:
    实体数据库,所述实体数据库中针对一个实体的记录包括实体数据字段和可变属性字段,在所述实体数据字段中存储有表示实体的实体数据,在所述可变属性字段中存储有描述所述实体的属性的实体属性数据;以及
    关系数据库,所述关系数据库中的每条记录包括两个节点和边信息,其中,在所述两个节点中分别存储有分别表示两个实体的两个实体数据,在所述边信息中存储有表示所述两个实体之间的关系的实体间关系数据。
  4. 根据权利要求3所述的方法,其中,
    所述实体数据库中针对一个实体的记录还包括元信息字段,在所述元信息字段中存储有与所述实体相关的元信息,所述元信息是使所述实体区别于其他实体的信息,
    在所述执行查询操作的步骤中,基于所述元信息来确定所述实体数据。
  5. 根据权利要求4所述的方法,其中,
    所述数据库还包括类别数据库,在所述类别数据库中,对应地存储有 多个实体类别数据和类别标签,所述多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据,
    在所述实体数据库中针对所述实体的记录中的元信息字段中,存储有与描述所述实体的类别的实体类别数据对应的类别标签,
    在执行根据所述实体反向查询语句的查询操作时,在所述实体数据库中检索满足下述条件的实体数据:
    类别标签为表示所述指定类别的实体类别数据或者从属于表示所述指定类别的实体类别数据所对应的类别标签;并且
    相应的实体属性数据满足所述指定属性过滤条件。
  6. 根据权利要求5所述的方法,其中,
    在所述类别数据库中,与每个实体类别数据关联地存储有针对该实体类别数据所表示的实体类别定义的实体属性,
    所述执行根据实体反向查询语句的查询操作的步骤包括:
    在所述指定属性过滤条件所涉及的指定属性属于为所述指定类别定义的实体属性的情况下,针对所述实体数据库执行所述根据实体反向查询语句的查询操作。
  7. 一种信息查询设备,包括:
    第一转换装置,用于将用户输入的自然语言查询文本转换为结构化的抽象查询语言语句;
    第二转换装置,用于将所述抽象查询语言语句转换为基本查询语句或迭代的多个基本查询语句;
    查询装置,用于针对预先准备的数据库,执行根据所述基本查询语句或所述迭代的多个基本查询语句的查询操作。
  8. 根据权利要求7所述的设备,其中,所述查询装置包括:
    实体信息查询装置,用于执行实体信息查询语句,以查询与指定实体相关的信息;
    实体属性查询装置,用于执行实体属性查询语句,以查询指定实体的指定属性;
    实体反向查询装置,用于执行实体反向查询语句,以根据指定属性过 滤条件和指定类别来反向查询对应的实体;
    相关实体查询装置,用于执行相关实体查询语句,以查询与指定实体具有指定关系的实体;以及
    实体间关系查询装置,用于执行实体间关系查询语句,以查询两个指定实体之间的关系。
  9. 根据权利要求8所述的设备,其中,
    所述数据库包括:
    实体数据库,所述实体数据库中针对一个实体的记录包括实体数据字段和可变属性字段、元信息字段,在所述实体数据字段中存储有表示实体的实体数据,在所述可变属性字段中存储有描述所述实体的属性的实体属性数据,在所述元信息字段中存储有与所述实体相关的元信息,所述元信息是使所述实体区别于其他实体的信息,所述查询装置基于所述元信息来确定所述实体数据;
    关系数据库,所述关系数据库中的每条记录包括两个节点和边信息,其中,在所述两个节点中分别存储有分别表示两个实体的两个实体数据,在所述边信息中存储有表示所述两个实体之间的关系的实体间关系数据;以及
    所述数据库还包括类别数据库,在所述类别数据库中,对应地存储有多个实体类别数据和类别标签,所述多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据,
    在所述实体数据库中针对所述实体的记录中的元信息字段中,存储有与描述所述实体的类别的实体类别数据对应的类别标签,
    其中,所述实体反向查询装置在所述实体数据库中检索满足下述条件的实体数据:
    类别标签为表示所述指定类别的实体类别数据或者从属于表示所述指定类别的实体类别数据所对应的类别标签;并且
    相应的实体属性数据满足所述指定属性过滤条件。
  10. 根据权利要求9所述的设备,其中,
    在所述类别数据库中,与每个实体类别数据关联地存储有针对该实体 类别数据所表示的实体类别定义的实体属性,
    所述实体反向查询装置在所述指定属性过滤条件所涉及的指定属性属于为所述指定类别定义的实体属性的情况下,针对所述实体数据库执行所述根据实体反向查询语句的查询操作。
  11. 一种信息查询方法,包括:
    将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句;
    将所述查询语句转换为基本查询语句或迭代的多个基本查询语句;
    在预先准备的数据库中,根据所述基本查询语句或所述迭代的多个基本查询语句,进行查询操作,以确定所述自然语言查询文本所对应的实体的特征信息。
  12. 根据权利要求11所述的信息查询方法,其特征在于,所述将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句包括:
    对所述自然语言查询文本进行意图分析,确定所述自然语言查询文本的查询意图;
    去除所述自然语言查询文本中,与所述查询意图不相符的文字内容;
    将所述去除后的自然语言查询文本转化为所述结构化的查询语句。
  13. 根据权利要求11所述的信息查询方法,其特征在于,所述在预先准备的数据库中,根据所述基本查询语句或所述迭代的多个基本查询语句,进行查询操作包括:
    在预先准备的数据库中,查询指定实体的指定属性;
    或,在预先准备的数据库中,查询与指定实体相关的信息;
    或,在预先准备的数据库中,根据指定属性过滤条件和指定类别来反向查询对应的实体;
    或,在预先准备的数据库中,查询与指定实体具有指定关系的实体;
    或,在预先准备的数据库中,查询两个指定实体之间的关系。
  14. 根据权利要求11所述的信息查询方法,其特征在于,所述在预先准备的数据库中,根据所述基本查询语句或所述迭代的多个基本查询语句,进行查询操作包括:
    在所述实体数据库中,使用所述基本查询语句查询实体数据字段,和/或可变属性字段,在所述实体数据字段中存储有表示实体的实体数据,在所述可变属性字段中存储有描述所述实体的属性的实体属性数据;以及
    在所述关系数据库中,使用所述基本查询语句查询记录,每条记录包括两个节点和边信息,其中,在所述两个节点中分别存储有分别表示两个实体的两个实体数据,在所述边信息中存储有表示所述两个实体之间的关系的实体间关系数据。
  15. 根据权利要求14所述的信息查询方法,其特征在于,所述在所述实体数据库中,使用所述基本查询语句查询实体数据字段,和/或可变属性字段包括:
    实体数据库中,使用所述基本查询语句查询元信息字段,在所述元信息字段中存储有与所述实体相关的元信息,所述元信息是使所述实体区别于其他实体的信息。
  16. 根据权利要求14所述的信息查询方法,其特征在于,所述在预先准备的数据库中,根据所述基本查询语句或所述迭代的多个基本查询语句,进行查询操作还包括:
    在所述类别数据库中,使用所述基本查询语句查询实体类别数据和类别标签,所述多个实体类别数据被划分为多个层次,较低层次的实体类别数据从属于与其关联的较高层次的实体类别数据;
    在所述实体数据库中查询针对所述实体的记录中的元信息字段,所述元信息字段中存储有与描述所述实体的类别的实体类别数据对应的类别标签。
  17. 根据权利要求13所述的信息查询方法,其特征在于,在预先准备的数据库中,根据指定属性过滤条件和指定类别来反向查询对应的实体包括:
    查询类别标签为表示所述指定类别的实体类别数据,或者查询从属于表示所述指定类别的实体类别数据所对应的类别标签;相应的实体属性数据满足所述指定属性过滤条件。
  18. 根据权利要求13所述的信息查询方法,其特征在于,还包括:
    当在所述指定属性过滤条件所涉及的指定属性属于为所述指定类别定义的实体属性时,执行步骤所述在预先准备的数据库中,根据指定属性过滤条件和指定类别来反向查询对应的实体。
  19. 一种信息查询装置,包括:
    第一转换模块,用于将用户输入的自然语言查询文本进行意图分析,转换为结构化的查询语句;
    第二转换模块,将所述查询语句转换为基本查询语句或迭代的多个基本查询语句;
    查询模块,用于在预先准备的数据库中,根据所述基本查询语句或所述迭代的多个基本查询语句,进行查询操作,以确定所述自然语言查询文本所对应的实体的特征信息。
PCT/CN2016/070332 2015-02-13 2016-01-07 信息查询方法和设备 WO2016127740A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/550,591 US10860632B2 (en) 2015-02-13 2016-01-07 Information query method and device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510081199.X 2015-02-13
CN201510081199.XA CN104636478B (zh) 2015-02-13 2015-02-13 信息查询方法和设备

Publications (1)

Publication Number Publication Date
WO2016127740A1 true WO2016127740A1 (zh) 2016-08-18

Family

ID=53215224

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/070332 WO2016127740A1 (zh) 2015-02-13 2016-01-07 信息查询方法和设备

Country Status (3)

Country Link
US (1) US10860632B2 (zh)
CN (1) CN104636478B (zh)
WO (1) WO2016127740A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106776888A (zh) * 2016-11-30 2017-05-31 北京赛迈特锐医疗科技有限公司 智能结构化搜索系统及其搜索方法
EP3582121A4 (en) * 2017-01-10 2020-05-20 Alt Inc. COMPUTER SYSTEM, SERVER DEVICE AND PROGRAM FOR USING A SET OF ATTRIBUTED INFORMATION IN CONNECTION WITH AN INDIVIDUAL
CN112559690A (zh) * 2020-12-21 2021-03-26 广东珠江智联信息科技股份有限公司 一种自然语言智能数据建模技术
US11030247B2 (en) 2017-12-29 2021-06-08 Electronic Arts Inc. Layered graph data structure
CN114168726A (zh) * 2022-02-11 2022-03-11 阿里巴巴达摩院(杭州)科技有限公司 语句处理方法、装置、存储介质及系统

Families Citing this family (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104636478B (zh) 2015-02-13 2019-12-20 广州神马移动信息科技有限公司 信息查询方法和设备
CN106610973B (zh) * 2015-10-21 2020-09-04 北京国双科技有限公司 电视数据查询方法及装置
CN106339454A (zh) * 2016-08-25 2017-01-18 北京云知声信息技术有限公司 查询命令转化方法及装置
CN106951452A (zh) * 2017-02-22 2017-07-14 中国建设银行股份有限公司 用于账务追溯的数据查询方法及装置
CN107357812A (zh) * 2017-05-31 2017-11-17 阿里巴巴集团控股有限公司 一种数据查询方法及装置
CN107480183A (zh) * 2017-07-06 2017-12-15 广州索答信息科技有限公司 一种厨房领域知识图谱查询方法、装置和存储介质
US10990592B2 (en) * 2017-10-31 2021-04-27 Microsoft Technology Licensing, Llc Querying of profile data by reducing unnecessary downstream calls
US11416481B2 (en) * 2018-05-02 2022-08-16 Sap Se Search query generation using branching process for database queries
CN110858100B (zh) * 2018-08-22 2023-10-20 北京搜狗科技发展有限公司 联想候选词生成方法及装置
CN109299129A (zh) * 2018-09-05 2019-02-01 深圳壹账通智能科技有限公司 自然语言的数据查询方法、装置、计算机设备及存储介质
CN109753658B (zh) * 2018-12-29 2023-09-19 百度在线网络技术(北京)有限公司 交互方法和装置
CN109933672B (zh) * 2019-02-12 2021-09-07 北京百度网讯科技有限公司 处理查询的方法、装置、电子设备和计算机可读存储介质
CN111538743B (zh) * 2020-04-22 2023-08-18 电子科技大学 基于sql的数据血缘关系分析方法以及系统
CN111897829A (zh) * 2020-05-26 2020-11-06 华瑞新智科技(北京)有限公司 一种用于医疗软件的自然语言查询方法及设备
US11604790B2 (en) * 2020-08-31 2023-03-14 Unscrambl Inc Conversational interface for generating and executing controlled natural language queries on a relational database
US20220067102A1 (en) * 2020-09-03 2022-03-03 International Business Machines Corporation Reasoning based natural language interpretation
CN112507098B (zh) * 2020-12-18 2022-01-28 北京百度网讯科技有限公司 问题处理方法、装置、电子设备、存储介质及程序产品
CN112632106B (zh) * 2020-12-29 2023-05-23 重庆农村商业银行股份有限公司 一种知识图谱查询方法、装置、设备及存储介质
CN112632110A (zh) * 2021-01-06 2021-04-09 上海合合信息科技股份有限公司 一种自动生成查询语句的数据提取方法及装置
CN113239009A (zh) * 2021-04-08 2021-08-10 大唐软件技术股份有限公司 一种数据库操作方法、装置、设备及存储介质
CN113127506B (zh) * 2021-06-16 2021-10-15 北京明略昭辉科技有限公司 目标查询语句的构建方法、装置、存储介质和电子装置
CN114461665B (zh) * 2022-01-26 2023-01-24 北京百度网讯科技有限公司 用于生成语句转换模型的方法、装置及计算机程序产品

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050289124A1 (en) * 2004-06-29 2005-12-29 Matthias Kaiser Systems and methods for processing natural language queries
CN102262634A (zh) * 2010-05-24 2011-11-30 北京大学深圳研究生院 一种自动问答方法及系统
CN102737049A (zh) * 2011-04-11 2012-10-17 腾讯科技(深圳)有限公司 一种数据库的查询方法和系统
CN103646032A (zh) * 2013-11-11 2014-03-19 漆桂林 一种基于本体和受限自然语言处理的数据库查询方法
CN104636478A (zh) * 2015-02-13 2015-05-20 广州神马移动信息科技有限公司 信息查询方法和设备

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7461052B2 (en) * 2004-12-06 2008-12-02 International Business Machines Corporation Abstract query plan
US7333981B2 (en) * 2004-12-17 2008-02-19 International Business Machines Corporation Transformation of a physical query into an abstract query
US20080016049A1 (en) * 2006-07-12 2008-01-17 Dettinger Richard D Natural language support for query results
US7680780B2 (en) * 2007-04-06 2010-03-16 International Business Machines Corporation Techniques for processing data from a multilingual database
WO2012047530A1 (en) * 2010-09-28 2012-04-12 International Business Machines Corporation Providing answers to questions using logical synthesis of candidate answers
US9646164B2 (en) * 2010-12-30 2017-05-09 Aziomatics Ab System and method for evaluating a reverse query
CN102682118B (zh) * 2012-05-15 2015-02-04 北京久其软件股份有限公司 一种多维数据模型访问方法及装置
CN102737134B (zh) * 2012-06-29 2014-06-18 电子科技大学 适用于大规模实时数据流的查询处理方法
CN102982147B (zh) * 2012-11-26 2016-03-30 深圳市华为技术软件有限公司 一种提高数据信息整合效率的方法和装置
CN103440296A (zh) * 2013-08-19 2013-12-11 曙光信息产业股份有限公司 数据查询方法和装置
US20170116260A1 (en) * 2014-02-25 2017-04-27 Google Inc. Using a dimensional data model for transforming a natural language query to a structured language query

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050289124A1 (en) * 2004-06-29 2005-12-29 Matthias Kaiser Systems and methods for processing natural language queries
CN102262634A (zh) * 2010-05-24 2011-11-30 北京大学深圳研究生院 一种自动问答方法及系统
CN102737049A (zh) * 2011-04-11 2012-10-17 腾讯科技(深圳)有限公司 一种数据库的查询方法和系统
CN103646032A (zh) * 2013-11-11 2014-03-19 漆桂林 一种基于本体和受限自然语言处理的数据库查询方法
CN104636478A (zh) * 2015-02-13 2015-05-20 广州神马移动信息科技有限公司 信息查询方法和设备

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106776888A (zh) * 2016-11-30 2017-05-31 北京赛迈特锐医疗科技有限公司 智能结构化搜索系统及其搜索方法
EP3582121A4 (en) * 2017-01-10 2020-05-20 Alt Inc. COMPUTER SYSTEM, SERVER DEVICE AND PROGRAM FOR USING A SET OF ATTRIBUTED INFORMATION IN CONNECTION WITH AN INDIVIDUAL
US11030247B2 (en) 2017-12-29 2021-06-08 Electronic Arts Inc. Layered graph data structure
CN112559690A (zh) * 2020-12-21 2021-03-26 广东珠江智联信息科技股份有限公司 一种自然语言智能数据建模技术
CN114168726A (zh) * 2022-02-11 2022-03-11 阿里巴巴达摩院(杭州)科技有限公司 语句处理方法、装置、存储介质及系统
CN114168726B (zh) * 2022-02-11 2022-05-31 阿里巴巴达摩院(杭州)科技有限公司 语句处理方法、装置、存储介质及系统

Also Published As

Publication number Publication date
CN104636478A (zh) 2015-05-20
CN104636478B (zh) 2019-12-20
US20180039692A1 (en) 2018-02-08
US10860632B2 (en) 2020-12-08

Similar Documents

Publication Publication Date Title
WO2016127740A1 (zh) 信息查询方法和设备
JP4644420B2 (ja) ネットワークを介してデータを検索及び提示する方法及びマシン可読記憶装置
CN111291161A (zh) 法律案件知识图谱查询方法、装置、设备及存储介质
JP6014725B2 (ja) 単文/複文構造の自然言語クエリに対する検索および情報提供方法並びにシステム
WO2021174783A1 (zh) 近义词推送方法、装置、电子设备及介质
Ngonga Ngomo On link discovery using a hybrid approach
Gilson et al. From web data to visualization via ontology mapping
US20130262449A1 (en) System and method for search refinement using knowledge model
US20130138627A1 (en) Quick Find for Data Fields
US20090112903A1 (en) Ontology data import/export method and apparatus
WO2022001682A1 (zh) 一种车载系统的控件对象查询方法和装置
Jin et al. Querying business process models based on semantics
JP2019040598A5 (zh)
US20230205996A1 (en) Automatic Synonyms Using Word Embedding and Word Similarity Models
Matei et al. OLAP for multidimensional semantic web databases
CN113010660A (zh) 一种基于知识图谱的智能问答方法及装置
CN105989097A (zh) 一种基于本体的知识库查询方法及系统
CN110321446B (zh) 相关数据推荐方法、装置、计算机设备及存储介质
KR20120047622A (ko) 디지털 콘텐츠 관리 시스템 및 방법
Barbella et al. A semi-automatic data integration process of heterogeneous databases
Soylu et al. Towards exploiting query history for adaptive ontology-based visual query formulation
Chen et al. Trends in conceptual modeling: Citation analysis of the ER conference papers (1979-2005)
Song et al. Semantic query graph based SPARQL generation from natural language questions
CN108804580B (zh) 一种在联邦型rdf数据库中查询关键字的方法
Modoni et al. A semantic framework for graph-based enterprise search

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16748541

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15550591

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16748541

Country of ref document: EP

Kind code of ref document: A1