JP2017204157A - Chat system, management server and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a chat service by a bot even to a client without a technology of the bot.SOLUTION: A chat system has: a management server; a client server; a chat application server; and a plurality of user terminals. The management server has: a communication unit that communicates with a user database in which attribute information on users is stored, a content database in which contents to be distributed to the user terminal are stored, and a client server; a submission sentence generation unit that generates submission sentences, and transmits the generated submission sentence to the client server; an analysis unit that implements natural language processing with respect to return sentences to be transmitted from the user terminal, acquires the user attribute information and evaluation information, and registers the acquired attribute information and evaluation information in the user database; a user preference property learning unit that learns a preference property of the user on the basis of the user attribute information stored in the user database; and a content selection unit that selects the content to be distributed to the user from the content database on the basis of a learning result of the user preference property learning unit.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a chat system, a management server, and a program that collect user preferences and provide content that matches the preferences.

  Conventionally, as a recommended service technology, there is a technology that distributes information about a product or service to be recommended to a user based on purchase history information of a user's product or service, viewing / browsing history information, or the like.

  In addition, there is a technique for returning weather information or the like of a place requested by the user as a posted message from the content of the posted message transmitted from the user (for example, Non-Patent Document 1). Here, the posted text to be returned to the user is automatically posted by the bot technique. For example, according to a representative Twitter (registered trademark) (asynchronous chat system) as an SNS service, a post like a real person can be periodically and automatically transmitted using bot. it can.

"Weather bot", [online], [Search August 30, 2011], Internet <URL: http://twitter.com/otenki_bot/>

  However, when chatting using a bot, an appropriate conversation sentence (posted sentence) cannot be generated unless the user's attribute is known, and it is important to know the user's attribute.

  Furthermore, sentence generation using the bot technology is sophisticated, and a small client cannot easily use chat using a bot.

  Therefore, the present invention provides a chat system, a management server, and a program capable of providing a chat service using a bot even if the client does not have the bot technique in interactive communication with a user in the chat service. For the purpose.

  One aspect of the present invention is a chat system, which includes a management server, a client server, a chat application server, and a plurality of user terminals, and the management server stores a user database in which user attribute information is stored. A content database that stores content to be distributed to the user terminal, a communication unit that communicates with the client server, a posted message generation unit that generates a posted message and transmits the posted message to the client server, and the user terminal Based on the user attribute information stored in the user database and the analysis unit that performs natural language processing on the sent reply sentence, acquires user attribute information and evaluation information, and registers in the user database , A user preference learning unit that learns a user preference and a learning result of the user preference learning unit Based on a chat system and a content selection unit that selects content to be delivered to the user from the content database.

  One aspect of the present invention is a chat server management server that communicates with a client database, a user database that stores user attribute information, a content database that stores content to be distributed to a chat user's user terminal, and a client server. A communication unit to perform, a posted message generation unit that generates and transmits a post message to the client server, and natural language processing is performed on a reply message transmitted from the user terminal, and user attribute information and evaluation information are obtained. An analysis unit that acquires and registers in the user database, a user preference learning unit that learns a user preference based on user attribute information stored in the user database, and learning of the user preference learning unit Based on the result, a container for selecting content to be distributed to the user from the content database. A management server and a tool selection unit.

  One aspect of the present invention is a program for a chat system, which generates a posted sentence and transmits a posted sentence to the client server, and a natural language process for a reply sentence transmitted from the user terminal. The user attribute information and the evaluation information, and the user's preference based on the analysis process registered in the user database in which the user attribute information is stored and the user attribute information stored in the user database. User preference learning processing to be learned, and content selection processing to select content to be delivered to the user from a content database in which the content to be delivered to the chat user's user terminal is stored based on the learning result of the user preference learning processing Is a program that causes a computer to execute.

  The present invention can provide a chat service by a bot even for a client without a bot technology, and further, by collecting user attributes and automatically learning user preferences, content suitable for users Can be delivered.

FIG. 1 is a system configuration diagram according to an embodiment of the present invention. FIG. 2 is a block diagram of the management server 1 in this embodiment. FIG. 3 is an operation sequence diagram according to the embodiment of the present invention.

  Embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a system configuration diagram according to an embodiment of the present invention.

  According to FIG. 1, a management server 1, a client server 2, a chat application server 3, and a plurality of user terminals 4 are connected to the Internet.

  Here, the chat application server 3 is, for example, a LINE (registered trademark) application server, and the user terminal 4 is a communication terminal such as a smartphone.

  According to FIG. 1, the user terminals 4 of a large number of users B and C connect to the chat application server 3 via the Internet. For example, a posted sentence of the user A (account A) can be sent to a plurality of other users B and C. (Accounts B and C) browse and make a reply (answer), and users A, B and C enjoy a conversation on the network like a chat.

  The client server 2 is a server that distributes a question sentence or the like to the users B and C via the chat application server 3 as a recommendation service in interactive communication with the user in the chat service, and is an account A. . The client server 2 is a server of a service provider such as a service industry, for example.

  The management server 1 generates a question sentence to be distributed by the bot for the client server 2, and transmits the text of the generated question sentence to the client server 2. In addition, the answer sentences from the user terminals 4 of the users B and C are received via the client server 2, and the answer sentences are analyzed by natural language processing to analyze the user's preference. Furthermore, the optimum content for the user is selected based on the analyzed user preferences.

  FIG. 2 is a block diagram of the management server 1 in the present embodiment.

  The management server 1 includes a posted sentence generation unit 10, a natural language processing unit 11, a user attribute registration unit 12, a user preference learning unit 13, a content selection unit 14, a content distribution unit 15, and a user database 16. And a content database 17.

  The posted sentence generation unit 10 uses a template sentence to generate a question sentence for extracting profiles (user attributes) of a plurality of other users B and C (accounts B and C), a content distribution sentence, and the like. , Has the function of Bot. The generated text such as a question sentence is transmitted to the client server 2.

  The natural language processing unit 11 performs morphological analysis and dependency analysis on the response sentences and content evaluation sentences of the users B and C received by the client server 2.

  The user attribute registration unit 12 registers, in the user database 16, data necessary for determining the user's preference among the answer sentences and the evaluation sentences that have been subjected to natural language processing.

  The user preference learning unit 13 performs AI engine (Azure Machine Learning, Tensorflow, etc.) based on the attribute information of the user registered in the user database 16 and the content of the content database 17 in which the content given from the client is registered. Use) to learn user preferences. The AI engine that has finished learning, for example, automatically returns the result “BBB content should be issued” in response to the question “What is the best content for user B when he wants to obtain an AAA rating?” Output.

  Based on the output from the user preference learning unit 13, the content selection unit 14 selects content to be distributed, that is, content that matches the user preference.

  The content distribution unit 15 transmits the content selected by the content selection unit 14 to the client server 2.

  Next, the operation in the embodiment of the present invention will be described.

  FIG. 3 is an operation sequence diagram according to the embodiment of the present invention.

  First, the management server 1 generates a question text for extracting a profile for a poster (S1) and transmits it to the client server 2. For example, a question sentence such as “How old are you? Where are you living? What is your occupation? What content are you interested in?” Is generated and transmitted to the client server 2.

  The client server 2 generates a question sentence by the chat API from the question sentence text received from the management server 1 (S2), and transmits it to the chat application server 3. And the chat application server 3 transmits a question sentence to the user terminal 4 (S3).

  The user B of the user terminal 4 browses the question text (S4), and generates and transmits an answer text about his / her profile to the chat application (S5). For example, I am 25 years old / I live in Shinjuku Ward / I am a company employee. / I'm interested in robots.

  The chat application server 3 that has received the response text transmits the response text to the client server 2 (S6). Then, the client server 2 that has received the response text transmits the response text to the management server 1 (S7).

  The management server 1 executes natural language processing of the received answer sentence (S8) and stores it in the user database (S9). For example, “age: 25, address: Shinjuku-ku, job: employee, interested: robots”. It should be noted that the analysis result may be displayed from the client server 2 after classifying the data stored in the user database.

  Next, based on the data stored in the user database, the management server 1 selects content suitable for each user (in this case, user B) from the content database (S10). Then, the selected content is transmitted to the client server 2. For example, since the chat application user was impressed with the robot, articles related to the robot were distributed, and so on.

  The client server 2 that has received the content transmits the content to the chat application server 3 (S11).

  The chat server 3 that has received the content transmits the content to the user terminal 4 (S12).

  The chat application user of the user terminal 4 browses the transmitted content (S13), evaluates the content, generates an evaluation sentence (S14), and transmits it to the chat application server. For example, because I am interested in specific processing such as robot programming, the content is abstract and not enough.

  The chat application server 3 transmits the evaluation content transmitted by the chat application user to the client server 2 (S15).

  The client server 2 that has received the evaluation statement transmits the evaluation statement to the management server 1 (S16).

  The management server 1 executes natural language processing of the received evaluation sentence and analyzes it (S17), and extracts keywords. Then, based on the extracted keyword, a user attribute is added to the user database (S18). For example, evaluation: 1, interested: "concrete processing", not_like: "abstract" (evaluation value is 0 if the article is rated good, 1 if it is bad, etc.).

  On the other hand, when the feedback of the evaluation sentence does not come from the user (S19), a feedback request text is generated (S20). For example, the chat application user requests comments and opinions about the article, such as “How was the article?” Or “Is it interesting?” Then, the evaluation statement request is transmitted to the client server 2.

  The client server 2 generates an evaluation request sentence by the chat API from the evaluation sentence request received from the management server 1, and transmits it to the chat application server 3 (S21). Then, the chat application server 3 transmits an evaluation request text to the user terminal 4 (S22). The user terminal 4 that has received the evaluation request text performs the processes of S13 and S14.

  The management server 1 learns the user preference based on the user attributes stored in the user database (S23), and transmits the result to the client server 2 in response to the request.

  The client server 2 receives the user preference result (S24), and displays the result.

  As described above in detail, according to the present embodiment, even a client without bot technology can provide a bot chat service, and further collect user attributes to collect user attributes. Since the preference is automatically learned, content suitable for the user can be distributed. The client server does not need to be provided with a special device for bot.

  In addition, the user attributes can be viewed from the client, and the trend of user preference can be confirmed.

  In the above-described embodiment, each unit is configured by hardware, but may be configured by a program that causes the information processing apparatus (CPU) to perform the above-described operation processing.

  Although the present invention has been described with reference to the preferred embodiments, the present invention is not necessarily limited to the above-described embodiments, and various modifications can be made within the scope of the technical idea.

DESCRIPTION OF SYMBOLS 1 Management server 2 Client server 3 Chat application server 4 User terminal 10 Posted sentence generation part 11 Natural language processing part 12 User attribute registration part 13 User preference learning part 14 Content selection part 15 Content distribution part 16 User database 17 Content database

Claims (3)

A chat system,
A management server, a client server, a chat application server, and a plurality of user terminals;
The management server
A user database in which user attribute information is stored;
A content database storing content to be distributed to the user terminal;
A communication unit for communicating with the client server;
A posted message generation unit that generates a posted message and sends it to the client server;
An analysis unit that performs natural language processing on a reply sentence transmitted from the user terminal, acquires user attribute information and evaluation information, and registers the information in the user database;
Based on user attribute information stored in the user database, a user preference learning unit that learns user preferences,
The chat system which has a content selection part which selects the content delivered to a user from the said content database based on the learning result of the said user preference learning part. A chat system management server,
A user database in which user attribute information is stored;
A content database in which content to be distributed to chat user terminals is stored;
A communication unit that communicates with the client server;
A posted message generation unit that generates a posted message and sends it to the client server;
An analysis unit that performs natural language processing on a reply sentence transmitted from the user terminal, acquires user attribute information and evaluation information, and registers the information in the user database;
Based on user attribute information stored in the user database, a user preference learning unit that learns user preferences,
A management server comprising: a content selection unit that selects content to be distributed to a user from the content database based on a learning result of the user preference learning unit. A chat system program,
A post message generation process for generating a post message and sending it to the client server;
An analysis process for performing natural language processing on a reply sentence transmitted from the user terminal, obtaining user attribute information and evaluation information, and registering in a user database in which the user attribute information is stored;
A user preference learning process for learning a user preference based on user attribute information stored in the user database;
A program for causing a computer to execute content selection processing for selecting content to be distributed to a user from a content database in which content to be distributed to a user terminal of a chat user is stored based on a learning result of the user preference learning processing.