JP7199000B2 - Dialogue management server, dialogue management method, and program - Google Patents

Description

The present invention relates to a dialogue management server, a dialogue management method, and a program.

2. Description of the Related Art Conventionally, as one of low - cost CS ( Customer Support) means in services provided by companies to customers on networks, a method of automatically responding to a conversation using a chat network or the like is known.

For example, Patent Literature 1 describes a chatbot system for providing a chat service over a network that responds to multiple users (Patent Literature 1).

JP-A-2009-3533

In the conventional system described in Patent Document 1, based on the text information corresponding to the question input by the user, an answer that matches the question is generated and sent to the user. However, the analysis of text information requires complicated information processing such as morphological analysis and syntactic analysis, and the accuracy of the analysis is not high, so there is a possibility that appropriate dialogue cannot be carried out.

Accordingly, several aspects of the present invention have been made in view of such circumstances, and can reduce the load related to information processing and eliminate ambiguity of the user's intention, thereby realizing more appropriate dialogue. It is an object of the present invention to provide a dialogue management server, a dialogue management method, and a program that can achieve a goal.

A dialogue management server according to an aspect of the present invention is a dialogue management server that responds to questions from one or more terminal devices, comprising: a reception unit that receives input of the questions from the terminal devices; an answer generation unit that generates a first answer to the question, the reception unit further receives an input of a stamp, which is image information, and the answer generation unit generates a specific meaning pre-associated with the stamp. Based on this, a second answer is generated for the input stamp.

a dialogue result management unit for storing, in the dialogue management server, dialogue results including the question, the first answer to the question, the input stamp, and the second answer to the stamp in a dialogue log database; , the response generation unit matches the question with the first answer and the second answer based on the specific meaning associated with the stamp stored in the dialog log database as the dialog result. weighting may be learned.

The dialogue management server further includes an output control unit that outputs the generated first answer, wherein the output control unit is more suitable as an answer to the question when multiple first answers to the question are generated. Control may be performed so that the first answer candidate group is preferentially output.

In the above dialogue management server, when the stamp associated in advance with the desire to output another answer candidate is input, the output control unit selects the second answer next to the first answer candidate group as the second answer. may be controlled so as to output the second answer candidate group that conforms to .

In the dialog management server, when the question and the first answer to the question are repeated a predetermined number of times with the terminal device, the request for voice interaction as a response to the first answer is provided in advance. The device may further include a voice interaction management unit that manages so as to enable voice interaction with the user when the associated stamp is input.

A dialogue management method according to an aspect of the present invention is a dialogue management method for answering questions from one or more terminal devices, comprising: receiving an input of the question from the terminal device; wherein the step of receiving further receives an input of a stamp, which is image information, and the step of generating is based on a specific meaning pre-associated with the stamp, A second reply is generated for the input stamp.

A program according to an aspect of the present invention provides a computer that responds to questions from one or more terminal devices, a receiving unit that receives input of the questions from the terminal devices, and a first answer to the input questions. function as an answer generating unit for generating, the receiving unit further receives an input of a stamp, which is image information, and the answer generating unit receives an input based on a specific meaning preliminarily associated with the stamp A program for generating a second answer to said stamp that has been stamped.

In the present invention, the term "unit" does not simply mean physical means, but also includes the case where the functions of the "unit" are implemented by software. In addition, even if the function of one "section" or device is realized by two or more physical means or devices, the functions of two or more "sections" or devices are realized by one physical means or device. can be

According to the present invention, it is possible to reduce the load related to information processing and eliminate the ambiguity of the user's intention, thereby realizing a more appropriate dialogue.

1 is a schematic configuration diagram (system configuration diagram) of a dialogue management system according to an embodiment of the present invention; FIG. 1 is a conceptual diagram showing an example of processing for converting a document such as business manual data created by a general word processor or spreadsheet software into a structured document such as OOXML (Office Open XML); FIG. FIG. 4 is a conceptual diagram showing an example of processing for converting a structured document such as OOXML into a structured document of a specific format specific to this embodiment; 4 is a conceptual diagram showing an example of stamp information according to the embodiment of the present invention; FIG. 4 is a flowchart showing an example of dialogue management processing according to the embodiment of the present invention; It is a figure which shows an example of the screen of the display part of the user terminal device which concerns on embodiment of this invention. It is a figure which shows an example of the screen of the display part of the user terminal device which concerns on embodiment of this invention. It is a figure which shows an example of the screen of the display part of the user terminal device which concerns on embodiment of this invention. It is a figure which shows an example of the screen of the display part of the user terminal device which concerns on embodiment of this invention. It is a figure which shows an example of the screen of the display part of the user terminal device which concerns on embodiment of this invention. It is a figure which shows an example of the screen of the display part of the user terminal device which concerns on embodiment of this invention.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. The following embodiments are examples for explaining the present invention, and are not intended to limit the present invention only to the embodiments. Also, the present invention can be modified in various ways without departing from the gist thereof. Furthermore, in each drawing, the same reference numerals are given to the same constituent elements as much as possible, and redundant explanations are omitted.

<First Embodiment>
FIG. 1 is a schematic configuration diagram (system configuration diagram) of a dialogue management system according to an embodiment of the present invention. As shown in FIG. 1, the dialogue management system 100 illustratively comprises a dialogue management server device 1 (dialogue management server), a dialogue log database (DB) 4, a user terminal device 6, and an operator terminal device 8. It is

The dialogue management server device 1 is a server computer that manages dialogue on a predetermined network N, and automatically answers questions from one or more user terminal devices 6, for example. The interaction management server device 1 exhibits a server function by running a predetermined server program on the server computer. "Dialogue on the network N" means dialogue between the user and the dialogue management server device 1 via the network N, and dialogue between the operator and the dialogue management server device 1 via the network N. say. For example, "dialogue on network N" is a question input from the user terminal device 6 to the dialogue management server device 1 via the network N, an answer generated by the dialogue management server device 1 corresponding to the question, and , containing the user's response to the answer. Further, the "dialogue on the network N" is a question input from the operator terminal device 8 to the dialogue management server device 1 via the network N, an answer generated by the dialogue management server device 1 corresponding to the question, and , containing the operator's response to the answer.

The dialogue log DB 4 is a database that records the dialogue between the user and the dialogue management server device 1 and the dialogue between the operator and the dialogue management server device 1. For example, the user's terminal device A dialogue log (dialogue result) including the question input to the dialogue management server device 1 from 6, the answer generated by the dialogue management server device 1 corresponding to the question, and the user's response to the answer is recorded. Note that the dialogue log DB 4 may be included in the dialogue management server device 1 .

The user terminal device 6 and the operator terminal device 8 are devices that output the content of the dialogue, and are, for example, information output devices such as laptops or notebook computers. The user terminal device 6 and the operator terminal device 8 may be information output devices including mobile phones such as smart phones, tablet terminals, and the like.

The network N is, for example, a communication line or communication network related to information processing including the Internet. There is no particular limitation as long as it is configured to allow data transmission/reception.

The dialogue management server device 1 exemplarily includes an information processing unit 10 that executes dialogue management processing for managing dialogue on a predetermined network N, information necessary for the dialogue management processing, and information generated by the dialogue management processing. , a technical term dictionary DB 12 storing technical terms of the industry in which the document is used, and a synonym dictionary DB 13 storing synonyms of words.

For example, the information processing unit 10 functionally includes an acquisition unit 15, a data conversion unit 17, a natural language processing unit 21, a weight setting unit 23, an answer generation unit 27, a dialogue result management unit 29, an output control unit 31, a voice It includes a recognition processing unit 33 and a voice dialogue management unit 35 .

It should be noted that each of the above-described units of the information processing unit 10 can be realized, for example, by using a storage area such as a memory or a hard disk, or by causing a processor to execute a program stored in the storage area. Also, each of the DBs 12 and 13 of the dialogue management server device 1 can be implemented by being executed by a processor.

The acquisition unit 15 acquires, for example, a document to be answered. Here, "documents" mean, for example, text documents such as business manuals and Q&A collections used in business, documents created with word processor software, documents created with presentation software, and the like.

The data conversion unit 17 (conversion unit) converts each of the text data contained in a predetermined unit in the document based on the data of the document whose content has a hierarchical structure such as chapters, chapters, sections, paragraphs, and sub-items. is converted into a document in a structured format by associating one or a plurality of hierarchies leading to each predetermined unit with identifiable information and headings of one or a plurality of hierarchies leading to each unit. The predetermined unit may be, for example, an item positioned at the end (bottom layer) of the hierarchical structure. In that case, each of the text data included in the item located at the end of the hierarchical structure is treated as information that can identify one or more layers leading to each item and headings of one or more layers leading to each item. Convert to a document in a structured format in correspondence.

2 and 3, processing for converting documents such as business manuals and Q&A collections into structured documents of a specific format specific to this embodiment will be described. Here, documents such as business manual data created by a general word processor or spreadsheet software are converted into structured documents such as OOXML (Office Open XML) and then used in the dialogue management server device 1 of the present embodiment. An example of conversion to a structured document in a specific format that is However, when converting a document such as a business manual into a structured document of a specific format specific to this embodiment, it is not always necessary to go through OOXML, and various conversion processes can be employed.

FIG. 2 is a conceptual diagram showing an example of processing for converting a document such as business manual data created by a general word processor or spreadsheet software into a structured document such as OOXML (Office Open XML). As shown in FIG. 2, the data conversion unit 17 converts the business manual data in word files into a file aggregate described in OOXML (Office Open XML).

FIG. 3 is a conceptual diagram showing an example of processing for converting a structured document such as OOXML (Office Open XML) into a structured document of a specific format specific to this embodiment. As shown in FIG. 3, the data conversion unit 17 converts the business manual data converted into a file aggregate described in OOXML into a specific XML format. Specifically, the data conversion unit 17 extracts the work manual data having a hierarchical structure so that it can be referred to for each structural unit of the document (volume, chapter, section, section, sub-item, etc.) Each of the text data contained in the structural unit (item) located in the extracted structural unit (in FIG. 3, "Do . The document is converted into a document in a structured format in association with information related to the headings of each layer, and the document is recorded in the recording unit 11 as document information BI (character information), which will be described later.

Returning to FIG. 1, the reception unit 19 receives an input of a question sentence (question) from the user terminal device 6 (terminal device). The reception unit 19 also receives an input of a stamp, which will be described later, from the user terminal device 6 .

The natural language processing unit 21 reads, for example, character information in a document and character information included in an input question, performs morphological analysis, and extracts each word. "Morphological analysis" is one of the natural language processing of computers, and morphological analysis is performed based on information such as the grammar of the target language and the part of speech of words from natural language text data without grammatical information annotations. It is an analysis process that divides into units of , and determines the part of speech, etc. for each morpheme. Also, the natural language processing unit 21 may execute dependency analysis (syntactic analysis), that is, examining the modification relation of the sentence from the part-of-speech information.

The weighting setting unit 23 calculates the appearance frequency of words included in the character information in the document and the character information included in the input question, and sets the weight for each word based on the calculated word appearance frequency.

The answer generation unit 27 generates an answer to the input question text and stamp. For example, the answer generator 27 matches the input question text with the text data included in each item of the recorded document information, extracts a plurality of items related to the question text, and extracts each of the extracted items. An answer (first answer) including one or more layers leading to the item and their headings is generated and returned to the user as an answer to the question text. Also, the answer generation unit 27 generates an answer (second answer) to the input stamp based on a specific meaning pre-associated with the stamp. Furthermore, the answer generator 27 may learn weighting for matching between questions and answers based on dialogue results stored in the dialogue log database 4, which will be described later.

The dialogue result management unit 29 displays dialogue results including a question input by the user, an answer (first answer) corresponding to the question, a stamp input by the user, and an answer to the stamp (second answer). is stored in the dialogue log database 4. The dialog result management unit 29 stores, for example, voice information recognized by the voice recognition processing unit 33, which will be described later, in the dialog log database 4 as the dialog result.

The output control unit 31 displays the question text input by the user, the answer corresponding to the question text, the stamp input by the user, and the answer to the stamp on the display units of the user terminal device 6 and the operator terminal device 8. (not shown) (see FIGS. 6 to 11 to be described later).

The voice recognition processing unit 33 recognizes voice information of voice dialogue between the user and the operator.

The voice dialogue management unit 35 manages so as to enable voice dialogue with a certain user when a question and an answer to the question are repeated a predetermined number of times.

The recording unit 11 records the document converted into the structured format as document information BI. In addition, the recording unit 11 stores dialogue result information TI (dialogue result) including a question text input by the user, an answer corresponding to the question text, a stamp input by the user, and an answer to the stamp for each user. Record in association with Furthermore, the recording unit 11 may record stamp information SI, which is information input via the network N. FIG.

A “stamp” is image information for interactive input via the network N. FIG.

FIG. 4 is a conceptual diagram showing an example of stamp information according to the embodiment of the present invention. As shown in FIG. 4A, the stamp is an illustration of the user's emotions, intentions and intentions, or a message to be conveyed, and may include text. Further, as shown in FIG. 4B, the stamp includes an illustration in association with text information indicating a score (answer accuracy information) indicating the accuracy of the answer, for example, 100 points, 75 points, etc. good too.

The recording unit 11 records the stamp in association with a specific meaning. For example, the stamp ST is recorded in association with a specific meaning of "I would like to ask you a question again."

The input of the stamp may be performed not only by the user's terminal device 6 but also by the operator terminal device 8, and the stamp may be used as the reply generated by the reply generation unit 27. FIG.

(Dialogue management processing)
Dialog management processing according to an embodiment of the present invention will be described with reference to FIGS. 5 to 10. FIG. FIG. 5 is a flow chart showing an example of dialogue management processing according to an embodiment of the present invention.

(Step S1)
The acquisition unit 15 shown in FIG. 1 acquires documents such as business manuals and Q&A collections.

(Step S3)
As shown in FIG. 3, the data conversion unit 17 shown in FIG. A structured document in which each of the included text data is associated with information that can identify one or more layers leading to each structural unit and headings of one or more layers leading to each structural unit. Convert.

(Step S5)
The recording unit 11 shown in FIG. 1 records the document converted into the structured format as the document information BI.

(Step S7)
The reception unit 19 shown in FIG. 1 receives an input of a question from the user.

FIG. 6 is a diagram showing an example of the screen of the display unit of the user terminal device according to the embodiment of the present invention. As shown in FIG. 6, as displayed on the screen G of the user terminal device 6 shown in FIG. is the case valid?" is accepted. Note that the user U icon and the icon T corresponding to the dialogue management server device 1 shown in FIG. 6 are examples, and can be changed as appropriate.

(Step S9)
The answer generation unit 27 shown in FIG. 1 matches the input question sentence with the text data included in each item of the recorded document information, extracts a plurality of items related to the question sentence, and extracts the extracted items. An answer including one or a plurality of hierarchies leading to each item and its heading is generated and returned to the user as an answer to the question text.

As shown in FIG. 6, when a question T1 is input, the answer generator 27 shown in FIG. 1 generates an answer T3 that matches the input question T1 and presents it on the screen G. FIG. The answer generation process will be described in more detail below.

First, the natural language processing unit 21 shown in FIG. 1 reads, for example, the question T1 "Is it valid if I bring my pension book with my personal identification document and no address is written?" The natural language processing unit 19, for example, “identity verification documents”, “person”, “confirmation”, “documents”, “verification documents”, “identity verification”, “pension book”, “pension”, “notebook”, “ At least one or more words such as "bring", "address", "description", and "effective" are cut out. At this time, the natural language processing unit 21 may refer to the technical term dictionary DB 12 and the synonym dictionary DB 13 as necessary. The natural language processing unit 21 also refers to part-of-speech information based on the result of morphological analysis and executes syntactic analysis.

Next, the weighting setting unit 23 calculates the appearance frequency of the words included in the question T1, and sets the weighting for each word based on the calculated appearance frequency of the words. The weighting setting unit 23 uses the extracted "identity verification documents", "person", "confirmation", "documents", "verification documents", "identity verification", "pension handbook", "pension", and "notebook". , “Bring”, “Address”, “Description”, and “Valid” . . . The weighting setting unit 23 may, for example, set higher weights for words with a high appearance frequency in one or a plurality of documents. Also, the weighting setting unit 23 may set a higher weighting for words that appear less frequently in one or a plurality of documents. Furthermore, the weighting setting unit 23 may set a higher weighting for words that appear more frequently in shorter documents. Information indicating the set weighting for each word may be recorded as document information BI in the recording unit 11 .

The answer generation unit 27 generates the words "identity verification document", "person", "confirmation", " Documents", "confirmation documents", "identity verification", "pension book", "pension", "notebook", "bring", "address", "description", "effective", etc. Based on the weighting set by 23, sentences suitable as an answer to the input question T1 are extracted from the document information BI shown in FIG. The answer generator 27 may, for example, search for each structural unit (volume, chapter, section, section, sub-item, etc.) included in the document information BI, and extract a suitable sentence as an answer.

For example, based on the weight set for each word included in the question T1, the answer generator 27 extracts each text included in the structural unit positioned at the end of the hierarchical structure of the document data recorded as the document information BI. Among the data, text data containing more one or more words for which a higher weight is set is extracted as text data more suitable for question T1. Note that the matching process is not limited to the above, and various matching processes can be adopted.

Then, the answer generation unit 27 extracts text data more suitable to the question T1, for example, "(1) Management Chapter 1...", "(2) Currency exchange..." shown in FIG. An answer T3 is generated by adding a template such as "Display top three candidates! Which one should you look at in detail?"

According to this configuration, by generating an answer including a template, it is possible to have a natural conversation in a form close to conversation with a human being.

As shown in FIG. 6, the output control unit 31 shown in FIG. 1 controls so that when multiple answers to an input question are generated, an answer that is more suitable as an answer to the question is preferentially output. For example, the output control unit 31 selects the top three answer groups (first answer group) that are more suitable as the answer T3 to the question T1, "(1) Management Chapter 1...", "(2) Currency Exchange...", and "(3) Management Chapter 5..." may be controlled to be output in order from the top on the screen G of the display unit.

According to this configuration, the user can easily grasp an answer that is more suitable as an answer to a certain question, so the user's convenience is further improved.

Then, as shown in the selection result T5, on the screen G, when "(1) Management Chapter 1..." is selected in the answer T3, the output control unit 31 displays "( 1) It is configured to output the specific contents of "Management, Chapter 1...". Specifically, based on the selected number, the output control unit 31 converts the explanatory text of the sub-item in the constituent unit (volume, chapter, section, paragraph, sub-item, etc.) corresponding to the number to the document information BI. Extract from and output. Regarding the mode of selection in the answer T3, the number whose details you want to check in the answer T3 may be selected by clicking on the screen G, or the number whose details you want to check may be text-inputted in the entry field EB. You may choose.

(Step S11)
The recording unit 11 shown in FIG. 1 records stamps and specific meanings in association with each other.

Next, as shown in FIG. 6, when the user presses the stamp button SB on the screen G, the output control section 31 outputs a stamp selection screen SG1 including one or more stamps ST.

FIG. 7 is a diagram showing an example of the screen of the display unit of the user terminal device according to the embodiment of the present invention, showing a continuation of the content of the dialogue shown in FIG. On the stamp selection screen SG1 shown in FIG. 6, for example, when the user U selects the stamp ST1 by clicking or the like, the output control unit 31 outputs the stamp ST1 on the screen G as the user's response, as shown in FIG. do.

(Steps S13, 15)
The reception unit 19 shown in FIG. 1 receives step input from the user. For example, when the reception unit 19 receives the input of the stamp ST1, the answer generation unit 27 shown in FIG. (second answer) and presents the answer T9 on the screen G. Specifically, the answer generator 27 replies with an answer that prompts the user U to ask an additional question. In this way, the answer generating unit 27 does not execute complicated information processing such as morphological analysis and syntactic analysis as in the case where the text information "I would like to ask a question again" stamp ST1 is input, and the text information is It can generate an answer that is equivalent to what was entered.

FIG. 8 is a diagram showing an example of the screen of the display unit of the user terminal device according to the embodiment of the present invention, and is a diagram when the user asks an additional question. As shown in FIG. 8, when the question T11 "Is hospitalized ... necessary?" The top three answer candidate groups (first answer candidate group) that are more suitable as the answer T13 to the question T11, "(1) Management Chapter 1 ...", "(2) Management Chapter 22 ...", and "( 3) Create an answer T13 (first answer) including "Management Chapter 19..." and present it on the screen G. Next, when the user presses the stamp button SB on the screen G, the output control section 31 outputs a stamp selection screen SG3 including one or more stamps ST.

FIG. 9 is a diagram showing the continuation of the content of the dialogue shown in FIG. 8, and is a diagram in which an answer candidate group other than the first answer candidate group is newly output. For example, when the user U selects the stamp ST3 by clicking the stamp ST3 on the stamp selection screen SG3 shown in FIG. A stamp ST3 pre-associated with the desired meaning (specific meaning) is output on the screen G. Then, when the stamp ST3 is input, the answer generation unit 27 shown in FIG. 1 generates the first answer to the question T11 based on the specific meaning of "further details (desired to output other answer candidates)". Answer candidate group (second answer candidate group) that fits next to the first answer candidate group "(4) Deposit Chapter 4...", "(5) Automated Machine Card Chapter...", and "(6) Deposit Chapter 4 . . ” (second answer) is generated, and the output control unit 31 controls to output the answer T15.

Next, the dialogue result management unit 29 shown in FIG. 1 stores a question from the user, an answer to the question (first answer), an input stamp, and an answer to the input of the stamp (second answer). The dialog result including the dialog is stored in the dialog log database 4 . The processing of the dialogue result management unit 29 will be specifically described below.

FIG. 10 shows the user terminal device when the user selects the desired answer "(5) Automatic machine card version..." from the second answer group and enters a stamp including answer accuracy information indicating the accuracy of the answer. It is a figure which shows an example of the screen of a display part. As shown in FIG. 10, when the favorite button HI is selected after the stamp ST5 including the score (answer accuracy information) "100 points" indicating the accuracy of the answer is input, the dialogue result management unit 29 From the start to the end, that is, the text information of T1 to T17 and stamps ST1 to ST5 shown in FIGS. 6 to 10 are stored in the dialogue log database 4 as the dialogue result. For the user U, for example, the answer candidate group (second answer candidate group) "(4) Deposit Chapter 4...", "(5) Automatic Machine Card Chapter...", and "(6) Deposit Chapter 4..." The included answer "(5) Automated machine card version..." is an answer that is more suitable than the question T11 "Is hospitalized... necessary?" In this way, the dialogue result management unit 29 collectively manages the dialogue from the start to the end, so that the dialogue management server device 1 evaluates and presents an answer that is more suitable for the input question. Answers and answers actually selected by the user (desired by the user) are collectively managed.

Specifically, based on the stamp ST1 stored as the dialogue result, the dialogue result management unit 29 determines that the answer T7 before the stamp ST1 was input was inappropriate as an answer to the question T1 (answer accuracy was low). ).

Further, based on the stamp ST3 stored as the dialogue result, the dialogue result management unit 29 determines that although the first answer candidate group is suitable as an answer candidate for the question T11, the answer more suitable for the user to the question T11 is the second answer. It is analyzed that it existed in the candidate group (candidate display accuracy was low).

Further, the dialogue result management unit 29 stores the dialogue result including the stamp ST5, and classifies the dialogue result according to the answer accuracy based on the answer accuracy information. In other words, the dialogue result management unit 29 repeatedly records the dialogue result together with a stamp reflecting the accuracy of the input answer, and classifies and manages a plurality of dialogue results according to the accuracy of the answer.

Based on the specific meaning associated with the stamp stored in the dialogue log database 4 as the dialogue result, the answer generator 27 generates the input question, the answer to the question (first answer), and the answer to the stamp (first answer). 2nd answer) is learned.

For example, for the user U, the answer generation unit 27 may give the second answer candidate group output after the stamp ST3 is input rather than the answer T13 (first answer) corresponding to the first answer candidate group to the question T11. Analyzing that the answer "(5) Automatic machine card edition..." included in the corresponding answer T15 (second answer) is a more suitable answer, the weighting of the answer "(5) Automatic machine card edition..." Control so that it rises compared to the answer of For example, the answer generating unit 27 increases the weighting of the answer "(5) Automatic machine card edition...", and the answers "(1) Management chapter 1..." and "(2) Management chapter 22" included in the answer T13. …” and “(3) Chapter 19 of the management section …” may be weighted down. In this way, the answer generation unit 27 learns weighting for matching between input questions and answers by repeatedly learning based on dialogue results managed for each dialogue.

Further, when stamp ST5 (answer accuracy of 100 points) is included in the dialogue result, the answer generation unit 27 determines that the accuracy of the answer to the input question was good, and determines the answer before stamp ST5 was input. may be given more weight than other answers. On the other hand, if a stamp with a low answer accuracy is included, the previous answer in which the stamp was entered may be given a lower weight than the other answers.

As described above, according to the first embodiment of the present invention, in the dialogue between the user terminal device 6 and the dialogue management server device 1 via the network N, the information input by the user is associated with a specific meaning. Adopt a stamp. Therefore, there is no need to execute complicated information processing such as morphological analysis and syntactic analysis, which is required when text information is input. By grasping the , the ambiguity of the user's intention can be resolved, and a more appropriate dialogue can be realized. In addition, by analyzing the dialogue results that include stamps with specific meanings, it becomes easier to grasp the correctness of answers and dialogue trends.

In addition, the answer generator 27 learns weighting for matching between input questions and answers by repeatedly learning based on dialogue results managed for each dialogue.

According to this configuration, it is possible to learn the weighting of matching between questions and answers, so it is possible to improve the accuracy of answers.

Furthermore, the dialogue result management unit receives a stamp including a score indicating the accuracy of the answer as a response to the answer, and further generates a dialogue result including the input question text, the answer corresponding to the question text, and the stamp. The dialogue result is stored and classified according to the answer accuracy based on the answer accuracy.

According to this configuration, by classifying and managing the dialogue results by answer accuracy, it can be effectively used for classification in dialogue log big data analysis, and the analysis results can be used to optimize and improve the efficiency of business flows in companies. , it is possible to make it easier to use for employee education, etc.

<Second embodiment>
FIG. 11 is a diagram showing an example of the screen of the display unit of the terminal device when the question (or response) and the answer to the question (or response) are repeated a predetermined number of times. As shown in FIG. 11, the voice dialogue management unit 35 shown in FIG. When the stamp ST7 pre-associated with the user U is input, management is performed so that voice interaction with the user U becomes possible. For example, the voice dialogue management unit 35, as shown in the texts T11 to T15 and the stamp ST3 in FIG. Manage access to telephone lines so that operators can interact via telephone. Note that the predetermined number of times is not limited to two. It may be performed once, or may be performed three times or more.

In the case where the question and the answer are executed, for example, twice with the user U, and the user U makes a request for a phone call (input of a stamp ST7 indicating a desire for voice dialogue), the following is shown in FIG. The output control unit 31 may be configured to display on the screen G a message T21 for permitting the call.

Dialogue by telephone may be performed between the user terminal device 6 and the operator's operator terminal device 8 shown in FIG. may Voice information of the user and the operator is collected by the user terminal device 6, the operator terminal device 8, and other communication devices, and transmitted to the dialogue management server device 1. FIG. Note that the screen G may be displayed in common on the display unit of the user terminal device 6 and the display unit of the operator terminal device 8 . The operator can respond by telephone based on the dialogue on the screen G up to the dialogue by telephone.

Here, the speech recognition processing unit 33 of the dialogue management server device 1 shown in FIG. may manage the registration or update of voice information provided.

As described above, according to the second embodiment of the present invention, when a question and an answer to the question are repeated a predetermined number of times with a certain user, the voice interaction management unit 35 uses the voice as the answer to the answer. When the stamp ST7 indicating the desire for dialogue is input, management is performed so that voice dialogue is possible with the user. Therefore, by preventing the user from calling the operator easily, the number of calls to the operator can be reduced, thereby improving the efficiency of work within the organization.

It should be noted that each of the above-described embodiments is for facilitating understanding of the present invention, and is not intended to limit and interpret the present invention. The present invention may be modified/improved without departing from its spirit, and the present invention also includes equivalents thereof.

Each step S in the flow chart shown in FIG. 5 does not necessarily have to be executed exactly as described above. For example, steps S1, S3 and S5 may be performed after step S7. Further, step S11 may be executed at least before step S15, and may be executed before step S1 is started, for example, as a prerequisite for execution of the dialogue management process.

REFERENCE SIGNS LIST 1 dialogue management server device, 4 dialogue log DB, 6 user terminal device, 8 operator terminal device, 10 information processing section, 11 recording section, 12 technical term dictionary DB, 13 synonym dictionary DB, 15... Acquisition unit 17... Data conversion unit 19... Reception unit 21... Natural language processing unit 23... Weight setting unit 27... Answer generation unit 29... Dialogue result management unit 31... Output control unit 33... Speech recognition processing unit 35 Spoken dialogue management unit 100 Dialogue management system

Claims (6)

A dialogue management server that answers questions from one or more terminal devices,
a reception unit that receives input of the question from the terminal device, the reception unit that further receives input of a stamp that is image information;
an answer generation unit that generates a first answer to the input question, and an answer generation unit that generates a second answer to the input stamp based on a specific meaning previously associated with the stamp; ,
a dialog result management unit that stores, in a dialog log database, dialog results including the question, the first answer to the question, the input stamp, and the second answer to the stamp ;
The answer generation unit performs matching between the question and the first answer and the second answer based on the specific meaning associated with the stamp stored in the dialogue log database as the dialogue result. learning weights,
Dialogue management server. Further comprising an output control unit that outputs the generated first answer,
When a plurality of first answers to the question are generated, the output control unit controls to preferentially output a first answer candidate group that is more suitable as an answer to the question,
The interaction management server according to claim 1. When the stamp pre-associated with the desire to output another answer candidate is input, the output control unit sets the second answer as the second answer next to the first answer candidate group. control to output the candidate group,
3. The interaction management server according to claim 2. The stamp pre-associated with the fact that the question and the first answer to the question are repeated a predetermined number of times with the terminal device, and that a voice dialogue is desired as a response to the first answer. is input, further comprising a voice dialogue management unit that manages to enable voice dialogue with the user,
3. The interaction management server according to claim 1 or 2. A computer-implemented dialogue management method for answering questions from one or more terminal devices, comprising:
receiving an input of the question from the terminal device;
generating a first answer to the entered question;
a step of further receiving an input of a stamp, which is image information;
generating a second answer to the input stamp based on a particular meaning pre-associated with the stamp;
a step of storing a dialogue result including the question, the first answer to the question, the input stamp, and the second answer to the stamp in a dialogue log database;
learning weights for matching the question with the first answer and the second answer based on the specific meaning associated with the stamp stored in the dialogue log database as the dialogue result;
including,
Dialogue management method. a computer that answers questions from one or more terminals,
a reception unit that receives input of the question from the terminal device, the reception unit that further receives input of a stamp that is image information;
an answer generation unit that generates a first answer to the input question, the answer generation unit that generates a second answer to the input stamp based on a specific meaning pre-associated with the stamp;
a dialog result management unit that stores, in a dialog log database, dialog results including the question, the first answer to the question, the input stamp, and the second answer to the stamp;
It functions as
The answer generation unit performs matching between the question and the first answer and the second answer based on the specific meaning associated with the stamp stored in the dialogue log database as the dialogue result. learning weights,
program.

Images