JP2003058464A - Question-answer system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a question-answer system capable of providing a suitable answer to a customer's question or the like, storing and sorting the question and the answer and improving the accuracy of the answer. SOLUTION: In the question-answer system, a plurality of relevant questions are displayed on a question item input picture, and when there is no suitable question in the displayed questions (question group (b)), a question is inputted to a question input column (a), the contents of the inputted question are analyzed by natural language processing, the question is clarified by interacting with a questioner as required, and a plurality of relevant questions are selected and displayed as a question group b'. When there is not suitable question also in the question group b', a user is urged to leave the question in a message board, a BBS or a chat, and when an answer to the question is sent to the user, the user automatically or manually registers the question and the answer in a question/answer DB in a DB 2 to learn the answer and utilize the answer for succeeding questions.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for responding to a question sent from a user, and in particular, it analyzes a user's question using natural language processing and responds to each user in detail. The present invention relates to a question answering system capable of answering.

[0002]

2. Description of the Related Art With the spread of electronic commerce, websites that accept complaints from customers by e-mail are increasing. In this case, e-mail reply support software having an e-mail distribution function and an answer sentence creation function is used.
This is to reduce manpower required to deal with questions, complaints, etc. from customers. Frequently Asked Questions (FAQ), which are frequently asked questions and answers posted on the Web, are often used.

As the e-mail reply support software, "Kana Response" of US company Kana Communications is known. Also, in Japan, NTT Advanced Technology's "match mail" is known. The "match mail" is for supporting the creation of answer examples by analyzing the contents of electronic mail and distributing them to operators, or extracting frequently asked questions from the stored mail.

As a conventional technique relating to electronic mail response, Japanese Patent Laid-Open No. 2000-33 published on December 8, 2000.
9314 "Automatic response method, dialogue analysis method, response sentence generation method, device thereof, and medium storing the program thereof"
(Applicant: Nippon Telegraph and Telephone Corporation, inventor: Nobuyuki Omori and others). It monitors the number of users and the frequency of utterances, and when it is below a certain level, it generates a response sentence using the dialogue analysis database when it detects a question sentence as an interactive mode. The question sentence and its response sentence are collected from the utterance and are stored in the dialogue analysis database.

There are also systems that provide various information online. For example, online manuals for products, medical information, real estate information, etc. are provided on the Internet.

[0006]

However, in the above-mentioned conventional electronic mail reply support software, only the reply of the electronic mail is supported, and in order to automatically generate the replying electronic mail, the log or the like is analyzed and manually. There is a problem in that it is necessary to improve accuracy, and manpower for automatic response cannot be significantly reduced.

Further, the conventional online information providing system has a problem that the user cannot easily retrieve the necessary information when the keyword searched by the user is different from the keyword provided by the information providing source.

The present invention has been made in view of the above circumstances, and it is possible to provide an appropriate answer to a customer's question or the like, and to store and classify the question and the answer to improve the accuracy of the answer. The purpose is to provide a possible question answering system.

Another object of the present invention is to provide a question answering system capable of accurately outputting corresponding information by linking a keyword used for a customer's question and information of an information provider.

[0010]

DISCLOSURE OF THE INVENTION The present invention for solving the problems of the above conventional example is a question answering system for accepting a question from a user via the Internet and answering the question, in which the server asks the question. In the input screen, when a question is entered, the question content is processed in natural language, and interactive processing with the user is realized to clarify the question content, and as a result, multiple questions that are highly relevant to the user's question are displayed. It selects from the database to display, and when any one of the multiple questions is selected, the corresponding answer is displayed, and the user selects the appropriate question from the multiple questions provided through the interactive process. You can get answers easily.

According to the present invention, in the above question answering system, the server can display a plurality of typical questions on the question input screen, and when any one of the plurality of questions is selected, the corresponding answer can be displayed. In addition, when a question is entered, natural language processing of the question content is performed and multiple related questions are selected and displayed from the database, and when any of the multiple questions is selected, the corresponding answer is displayed. The user can select an appropriate question from a plurality of questions provided and easily obtain an answer.

According to the present invention, in the above-mentioned question answering system, the server selects and displays a plurality of highly relevant questions from the database, which are judged from the previous access log, and is displayed on the question input screen. When any one is selected, the corresponding answer can be displayed, and when a question is entered, natural language processing of the question contents is performed and multiple related questions are selected and displayed from the database. When any one of the questions is selected, the corresponding answer is displayed, and the user can easily obtain an answer by selecting an appropriate question from a plurality of provided related questions.

According to the present invention, in the above question answering system, the server displays a screen for login or question input,
It displays a message related to the question previously asked from the previous access log, and can show the affinity to the user.

The present invention is a question answering system that includes a server that receives a question from a user via the Internet and answers the question, and a database that stores the answer to the question. When a question is entered on the screen, the question content is processed in natural language,
In order to confirm the content of the question, a question is asked to the user to realize interactive processing, and as a result of the interactive processing, the user's question is displayed as an answer and multiple questions that are highly relevant to the user's question are selected from the database. The corresponding answer is displayed when any one of the plurality of questions is selected. The user receives the display of the corresponding answer through the interactive process, and selects the appropriate answer from the plurality of questions provided. You can select a question and get an answer easily.

According to the present invention, in the above question answering system, the server stores the relationship between the word on the question side and the word on the answer side in a database, and when a question is input, the question side is subjected to natural language processing and dialogue processing. Words of the answer side are searched, the word of the answer side that is related is searched, the database is searched using the word of the answer side, the user's question is displayed as an answer, and a plurality of words highly related to the user's question are displayed. The question is selected from the database and displayed. By storing the relationship between the word on the question side and the word on the answer side, a highly accurate answer to the question can be provided.

According to the present invention, in the above-mentioned question answering system, the server stores the words related to the question acquired in the course of the interactive processing in the database and uses them for the answers to the subsequent questions. Learning can improve the accuracy of the response.

According to the present invention, in the above-mentioned question answering system, the server provides a message board for leaving a question, an electronic bulletin board on which a question can be written and expects an answer, and a chat which directly interacts with the operator, and responds to those questions. Answers are additionally stored in a database and used for answers to subsequent questions, and the accuracy of responses can be improved by updating the database.

According to the present invention, in the question answering system, the server analyzes the words acquired in the course of the interactive processing with the user and the content of the question from the user, and collects information related to the user. You can make accurate user-related information and use it for marketing.

According to the present invention, in the above question answering system, the server performs natural language processing of the question contents to extract and manage words of high interest, and manages a plurality of questions of high relevance based on the interest level. The user can be provided with a plurality of questions that are highly related to each other depending on the degree of interest, and the accuracy of the response can be improved.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described with reference to the drawings. The question answering system according to the embodiment of the present invention displays a plurality of related questions on a question item input screen, and when there is no appropriate question among the plurality of questions (question group), the question is answered. When you enter a question in the input field, the content of the question is analyzed by natural language processing, if necessary, the question is clarified by interacting with the questioner, and multiple related questions are selected and displayed. If there is no question, the message board, BBS or chat is urged to leave the question, and when the user receives an answer to the question, the question and the answer are automatically or manually registered and learned. It is used for questions. As a result, the user can easily obtain the answer to the question, and the support side can improve the accuracy of the answer.

Further, in the question answering system according to the embodiment of the present invention, since the questioner is registered in advance, the data collected by the above-mentioned interactive processing (in the form of a combination of question keyword and answer keyword) is classified and personalized. The e-CRM (electric-Customer Re-store) that stores in the information and builds a long-term relationship between the customer and the company by carefully responding to the customer's needs through various information analyzes such as customer behavior analysis, complaint analysis, message analysis, etc.
relationship management).

A question answering system (present system) according to an embodiment of the present invention will be described with reference to FIG. Figure 1
FIG. 1 is a schematic configuration diagram of a question answering system according to an embodiment of the present invention. As shown in FIG. 1, this system includes a server 1, a database (DB) 2, an operation terminal 3 for inputting to the server 1, and a plurality of customer PCs (personal computers) connected to the server 1 via the Internet. )
4 and.

Each part of the system will be specifically described. The server 1 includes a Web function realizing means, a natural language processing function realizing means, and a knowledge base management function realizing means. Further, the knowledge base management function implementing means includes a question handling processing means, a dialogue processing means, an analysis processing means, a user management means, and a knowledge management means.

Here, the Web function realizing means is DB2.
In addition to making the information therein accessible on the Web, a process of making necessary information displayable on the customer PC 4 in response to access from the customer PC 4 via the Internet is performed. The natural language processing function realizing means is a means for performing a language analysis on an input question item, extracting a keyword of the question item, and performing a process of grasping the essence of the question. Natural language analysis is currently being researched, and the accuracy of analysis is improving day by day. The knowledge base management function implementing means is a main part of the present invention, and selects a plurality of highly relevant questions from the natural language of the analyzed questions and provides them to the user, and also analyzes the questions. By performing processing such as preparing questions and answers that are more relevant to each other, further improving the accuracy of question handling through dialogue with the questioner, analyzing the question items by analysis processing, and importing as customer information. , To enhance marketing information to the customer.

The question handling processing means in the knowledge base management function implementing means displays a question group with a high question frequency for a user who has accessed for the first time or a user who has accessed for the first time in a long time, or for a user who has previously asked a question. A question group related to the previous question item is displayed so that a question group highly relevant to the user is displayed on the access screen. Here, although the related question is displayed before the user inputs the question, it may be considered that the related question is not displayed until the question is input.

Further, when the user inputs a question item, the question handling process unit passes the question item to the natural language processing function implementing unit and receives the content analyzed by the natural language processing function implementing unit. The question handling processing means displays a question group having high relevance from the analysis result.

If the analysis contents are not clear, the question handling means activates the interactive processing means to enter the interactive mode, asks the user a question using the interactive processing means, and further asks the question if necessary according to the answer. Go to clarify user questions.

Further, the question handling processing means displays a message board according to an instruction from the user, and stores the question items described in the message board in the DB2. Message board, question written on BBS, question received by telephone / FAX, question received by individual e-mail, question received by chat, and among the corresponding answers, question / answer DB by operator's selection Be registered with.

Further, the question handling processing means is a BBS (Bulletin Board System: electronic bulletin board) according to an instruction from the user.
Is displayed. The BBSs are classified according to the content of the questions, and the user will access the BBSs of the desired classification. Further, the question answering processing means can enter a chat in which the user can directly interact with the operator according to an instruction from the user.

In order to clarify the question from the user, the interactive processing means asks the customer's PC 4 the question by the server 1 of the present system for the user's question according to the result of the natural language analysis. Natural language analysis is performed on the answer, and if the question content is still not clear, the question is repeated.

The analysis processing means performs customer behavior analysis, comment statistical analysis, complaint statistical analysis, message analysis, and access status statistical analysis. The customer behavior analysis is an analysis of a customer's trend from a question, for example, it is possible to know the customer's satisfaction. This customer behavior analysis is e-
It will be useful for CRM. In the comment statistical analysis, the question items and the response contents are analyzed and used as a type judgment material. Complaint statistical analysis is to analyze complaints, and to answer questions.
It also has a function to display a list of unanswered items. The message analysis is to analyze the questions and the like written on the message board to help organize the questions.

The access status statistical analysis means the number of visits, the number of visits, the degree of satisfaction, the degree of interest, the rate of visits to related products,
This is to analyze the degree of product attention. Note that the degree of interest is not only the degree of interest based on the number of accesses, but also the number of words included in the content of the question from the natural language analysis is a measure of the degree of interest. Frequent words will be classified as minor items under the major items in the questionnaire. This classification will be described later.

The user management means manages the authority of the user and performs processing corresponding to access from the authorized user. For example, an operator has a wider range of access authority than a general user.

The knowledge management means organizes the contents of the question and the contents of the response, stores them in a form that can be easily used in the DB 2, and periodically reviews the contents. Since the questions and answers are organized by the knowledge management means, the answers to the questions are automatically learned, and the more the system is used, the more accurate the response can be.

DB2 is a question / answer DB and product information D
B, a user DB, and an access log DB. The question / answer DB stores the answers to the accepted questions and the answers will increase as the number of questions from the user increases. The product information DB stores knowledge information about products and the sales status of the products. Regarding the sales status of products, the analysis results by the analysis processing means are also incorporated. Here, the product information DB is assumed on the assumption of a site that handles products, but it may be considered to be the service information DB if a product other than products, for example, a service is handled.

The user DB stores the information of the user who has accessed. At the time of the first access, since user registration is performed, basic data is acquired by the registration. After that, data regarding the user's orientation (including the degree of interest) is stored from the access log of the user. In addition, keywords are acquired from the question items, answers, and dialogue contents in the user DB and added to the user information. This processing is performed by the question handling processing means and the dialogue processing means. The access log DB stores access logs of users and operators. The analysis processing means performs various analyzes from this access log DB.

The operation terminal 3 is a terminal which is connected to the server 1 and operated by an operator who provides support. An answer to a question is input from the operation terminal 3, an input is made by BBS or chat, and the analysis result by the analysis processing means is displayed. The customer PC 4 accesses the server 1 via the Internet, inputs / searches for a question, and leaves the question on the message board if necessary. Further, the customer PC 4 can perform BBS and chat.

Next, the processing operation of this system is shown in FIG.
~ It demonstrates using FIG. FIG. 2 is an explanatory diagram showing a question item input screen, FIG. 3 is an explanatory diagram showing the screen after inputting the question items, and FIG. 4 is a flowchart diagram showing processing of the present system. First, pre-registered customer PC 4
When accessing the server 1 and inputting questions etc.,
The server 1 displays the screen as shown in FIG. 2 on the customer PC 4. Specifically, the question input field a and the number of questions is large,
A question group b in which a plurality of representative questions are displayed in a list is displayed (S1). If the user has previously asked a question, the question group b related to the question may be displayed. Note that the process S1 may be omitted in order to simplify the process.

Here, if the question to be input is in the question group b (Yes in step S2) before inputting the question, the customer obtains an answer by clicking on the question. Can be done (S3). If it is not in the question group b (No in step S2), the question is written in the question column a (S4).

When the question is written in the question column a and the "question" button is clicked, the server 1 analyzes the question sentence by the natural language processing function implementing means, and if the question is unclear, the mode is changed to the interactive mode. Then, the server 1 performs an interactive process of asking a question to the customer PC 4 (S5). Then, when the question content becomes clear, a question group b ′ in which a plurality of questions having a high relevance to the question are listed in order of high relevance is displayed as shown in FIG. 3 (S6).

The user refers to the question group b ', and if there is a corresponding question (Yes in step S7), clicks the question to obtain an answer (S8). However, if there is no question corresponding to the question group b '(process S
In the case of No. 7), click the message part instructing to "leave a question on the message board". Then, the message board is displayed (S9), and the user writes the question matter on the message board. The questions written on the message board are
It is temporarily stored in DB2 (S10).

With respect to the question content written on the message board, the support side inputs an answer to the question using the operation terminal 3. The entered answer is associated with the question D
It is stored in the question / answer DB of B2 and is transmitted from the server 1 to the customer PC 4 who asked the question by electronic mail.

If the user attaches importance to responsiveness, he / she enters into a chat which directly interacts with the operator instead of the message board. Also, instead of the message board, the user can select BB
It is also possible to select S. In BBS, the support side or a third party who knows the answer to the question will respond. Writing to the BBS on the support side is performed by the operation terminal 3.

When the operator answers the question written on the message board, the question and the answer are automatically D
It is stored in the question / answer DB of B2. After that, the question and the answer are analyzed by the analysis processing means, and become the material used for the new question. Also, telephone, personalized email,
The content responded by the chat is also registered in the question / answer DB of DB2 by the selection of the operator of the operation terminal 3 and used for a new question. It is desirable that the contents of the question / answer DB be maintained by the operator regularly organizing the information.

Further, since the access log and the question content by the user are stored in the access log DB of the DB 2 by the server 1, the items related to the previous question content are displayed at the second and subsequent user logins or the question input. You may let me. Specifically, "I was asked a question on xx month xx
Is that right? It is also possible to display a message such as "when a user logs in or input a new question, and display a question group related to the contents of the previous question.
This allows the user to obtain a group of highly relevant questions and their answers when asking a question related to the content of the previous question, and improve the accuracy of the answers in the system.

Next, the analysis processing of this system will be described. As the analysis processing, as described above, customer behavior analysis, comment statistical analysis, complaint statistical analysis, message analysis, access status statistical analysis, etc. are performed. In particular,
Comment statistics, complaint statistics, and customer behavior can be analyzed by classifying question items, and interest levels can be analyzed by subjecting question contents to natural language processing.

The degree of interest will be described below as an analysis process relating to the contents of the question. Specifically, the question items are classified into large items (e-mail, web, shop, etc.) in advance, and words under the question are automatically registered as small items below them. For example, if the user is interested in “price” in the question about the large item “mail”, “price” is set as a small item and the number of times is counted. In other words, the degree of interest is counted by the number of times of access and the number of words that became a topic.

As described above, in the analysis process, the question items are analyzed by using the degree of interest, the language analysis process is performed on the input question, and as a result, the question group having a high degree of interest for the related word is selected by the user. If it is provided to the user, the user can obtain the question group and the answer that are highly relevant from the viewpoint of the degree of interest, so that the accuracy in the system can be improved.

According to this system, when accessing the server 1 from the customer PC 4 and inputting a question, a user who has accessed a certain period from the first access or the previous access is sent to the website. Since a plurality of typical questions (question group b) are displayed among the questions, the representative question group b
There is an effect that you can easily get answers by selecting from.

According to the present system, when a user who has previously accessed, when inputting a question item, clicks on a question item previously asked or a question in the previous question group b to obtain an answer, Since the question group b ′ related to the question is displayed, the related question can be easily selected by selecting from the question group b ′. In this case, a message related to the previous question can be displayed from the previous access log to the user at the time of logging in or inputting the question to appeal the affinity to the user.

According to the present system, in order to clarify the question matters for the question from the user, the server 1 asks the user a question by an interactive process and clarify the question content by the answer from the user. Since it is possible to display a group of questions that are highly related to each other, the words acquired in the process of the dialogue processing are stored in the DB2 in relation to the question items and used for the dialogue processing. The accuracy can be improved.

According to this system, the user can ask questions b,
If there is no related question even when referring to b ', the question can be left on the message board, and the support side sends the answer to the user by e-mail or displays it on the Web, and the question on the message board. Question and answer D
Since it is registered in B2, the user can directly obtain an appropriate answer by e-mail or the like, and the support side can use the contents registered in the question / answer DB for the following questions. There is an effect that the accuracy of the response in the system can be improved.

According to the present system, when the user demands the responsiveness of the answer, the user can enter the chat which can directly interact with the operator, and the user can obtain the answer quickly. Further, according to the present system, the question can be written in the BBS by the selection, and the operator writes the answer in response to the question in the BBS. In addition, if the contents written in the BBS are registered in the question / answer DB, they can be used for subsequent questions.

According to this system, the analysis processing means analyzes the customer behavior, the comment statistics, the complaint statistics, the access status statistics and the like with respect to the question. Since high-ranking words are also managed, it is possible to provide the user with a question group having a high degree of interest and to improve the accuracy of the response in the system.

Next, an online manual service system (another system) according to another embodiment of the present invention will be described. Another system associates a keyword (question side keyword) extracted from a question item from a customer PC by a dialogue processing means and a natural language processing function implementing means with a keyword used in a manual stored in a database (manual side keyword). When memorized and there is a question, the keyword on the question side is extracted, the corresponding keyword on the manual side is searched, the online manual data is searched using it, and the corresponding manual description part is displayed and output on the customer PC. , It is possible to output a highly accurate answer to a question.

An online manual service system (another system) according to another embodiment of the present invention will be described with reference to FIG. FIG. 5 is a schematic configuration diagram of an online manual service system according to another embodiment of the present invention. Another system is, as shown in FIG. 5, a server 1, a database (DB) 2, and a server 1.
Maker PC5 (personal computer) connected to
And a plurality of customer PCs (personal computers) 4 connected to the server 1 via the Internet.

Each part of another system will be specifically described.
The server 1 includes a Web function realizing means, a natural language processing function realizing means, and a knowledge base management function realizing means. These means are similar to those described in FIG. Further, the knowledge base management function implementing means includes each means such as a question answering processing means, a dialogue processing means, an analysis processing means, a user management means, a knowledge management means, as in FIG.

The DB2 stores online manual data provided by each maker, and further, customer PC
The relationship is stored in order to connect the question items input from 4 and the online manual data. This relevance is, for example, “toner” (keyword on the manual side) in the manual of the printer product, but when the questioner inputs “ink” (keyword on the question side), when “toner” is indicated, The relationship between the side keyword "toner" and the question side keyword "ink" is stored. This relation is that the dialogue processing means and the natural language processing function realizing means extract the keyword of the question item and the keyword of the manual data to be the answer, and store them in association with each other.

The customer PC 4 is a user of the manufacturer's product,
The user ID or the like is input to access the server 1, the question regarding the product is input, and the search result of the online manual on the server 1 is received.

The maker PC 5 is a device corresponding to the BBS or chat realized by the question answering processing means of the server 1, and responds to the question or the like sent from the customer PC 4 to the BBS or the like. Further, the maker PC 5 uses the analysis processing means of the server 1 to analyze customer behavior, statistically analyze comments,
Receives the results of complaint statistical analysis, message analysis, access status statistical analysis, etc.

Here, the difference between this system and another system will be described. The server 1 stores the manual data of the product from the manufacturer and stores it in the DB 2. Customer PC 4
When you access the website of the server 1 and enter the assigned user ID, etc., the server 1
Is determined to be appropriate, and if the ID is appropriate, the online manual data of the corresponding product can be accessed. That is, the server 1 holds information that associates the user ID and the like input from the customer PC 4 with the online manual data of the product, and the online manual data of the corresponding product can be accessed by this information.

When the customer PC 4 inputs a question about the product on the Web screen that accesses the online manual data of the product, the server 1 analyzes the entered question by the natural language processing function implementing means, and the analysis result. By searching the DB2, if there is a description part of the relevant manual, the relevant part is displayed and output to the customer PC 4, and the question group related to the question item is also displayed and output.

The search of the DB2 will be described in detail. The question items are analyzed by the natural language processing function implementing means to extract the question side keywords, and the related manual side keywords stored in the DB2 are extracted. Then, the online manual data in DB2 is searched using the keyword on the manual side to extract the relevant manual description part. If the keyword on the question side and the keyboard on the manual side are the same, the online manual data may be searched directly by the keyword on the question side.

If the manual side keyword that is related to the extracted question side keyword cannot be retrieved, the server 1 inquires of the customer PC 4 about the question side keyword and extracts another question side keyword. To search for relevant manual keywords. The relevance between the keyword on the question side and the keyword on the manual side acquired in the search process is additionally stored in DB2.

The customer PC 4 can refer to the portion described in the relevant manual and, if necessary, can select a question from the displayed related question group and refer to the answer. Also,
If the customer does not get a satisfactory answer by the above operation, he / she inputs a question from another point of view to search for an appropriate answer, and if the customer does not get a satisfactory answer, he / she leaves the question on the message board. Maker PC with chat function
It is also possible to have a question and answer session with the operator of item 5.
The association between the question side keyword and the manual side keyword regarding the question and answer by the message board and chat can be selected by the operator of the maker PC 5 and stored in the DB 2.

As described above, in another system, the question-side keyword and the manual-side keyword are stored in association with each other, and the manual-side keyword related to the question-side keyword extracted from the question items by natural language analysis is searched and used. Since the online manual data is searched by using the online manual data, there is an effect that the gap between the question item and the manual data can be minimized and the accuracy of the answer to the question can be improved.

In this other system, the dialogue processing means and the natural language processing function realizing means can additionally store the relevance between the question side keyword and the manual side keyword by repeating the dialogue, so that the system is used. There is an effect that the accuracy of the answer can be improved.

In this other system, the information provided to the customer is explained as the online manual data, but the provided information may be medical information or real estate information.

[0069]

According to the present invention, there is provided a question answering system for accepting a question from a user via the Internet and answering the question, wherein the server displays the question content when the question is entered on the question input screen. Natural language processing is performed, and interactive processing with the user is realized in order to clarify the content of the question. As a result, a plurality of questions that are highly relevant to the user's question are selected and displayed from the database. When any one is selected, the corresponding answer is displayed, and the user can easily obtain an answer by selecting an appropriate question from a plurality of questions provided through interactive processing.

According to the present invention, the server displays a plurality of typical questions on the question input screen, and when any one of the plurality of questions is selected, the corresponding answer can be displayed and the question can be displayed. A question answering system that performs natural language processing of the question contents when a question is input and selects multiple related questions from the database to display, and displays the corresponding answer when any one of the multiple questions is selected Therefore, there is an effect that the user can select an appropriate question from a plurality of questions provided and easily obtain an answer.

According to the present invention, on the question input screen, the server selects and displays a plurality of highly relevant questions from the database judged from the previous access log, and selects any one of the plurality of questions. When the question is entered, the question content is processed by natural language and multiple related questions are selected and displayed from the database, and any one of the multiple questions is displayed. Since the question answering system displays the corresponding answer when is selected, the user can select an appropriate question from a plurality of highly relevant questions provided and easily obtain an answer. There is.

According to the present invention, the server uses the above-mentioned question answering system that displays a message regarding the content of a previous question from the previous access log on the screen for logging in or entering a question. There is an effect that can be appealed.

According to the present invention, there is provided a question answering system including a server that receives a question from a user via the Internet and answers the question, and a database that stores the answer to the question. When a question is entered on the input screen, the question content is processed in natural language, and in order to confirm the question content, a question is asked to the user to realize interactive processing. As a result of the interactive processing, the user's question is asked. In addition to displaying the answers to the questions, multiple questions that are highly relevant to the user's question are selected and displayed from the database, and when any one of the multiple questions is selected, the corresponding answer is displayed. There is an effect that a corresponding answer is displayed through and an appropriate question can be easily selected from a plurality of provided questions.

According to the present invention, the server stores the relation between the word on the question side and the word on the answer side in the database, and when the question is input, the word on the question side is extracted by natural language processing and dialogue processing. Then, the words on the answer side that are related to each other are searched, the database is searched using the words on the answer side, the answers to the user's questions are displayed, and multiple questions that are highly related to the user's questions are displayed from the database. Since the question answering system is selected and displayed, the relationship between the word on the question side and the word on the answer side is stored, which has the effect of providing a highly accurate answer to the question.

According to the present invention, the server stores the words related to the question acquired in the course of the dialogue processing in the database and uses them for answering the subsequent questions. Learning has the effect of improving the accuracy of the response.

According to the present invention, the server provides a message board for leaving a question, an electronic bulletin board in which a question can be written and expects an answer, and a chat for directly interacting with the operator, and the answer corresponding to the question. Since the question answering system described above is additionally stored in the database and used for answering subsequent questions, updating the database has the effect of improving the accuracy of the answer.

According to the present invention, the server is the question answering system, which analyzes the words acquired in the course of the interactive process with the user and the contents of the question from the user, and collects the information related to the user. There is an effect that user-related information can be accurately made and can be utilized for marketing.

According to the present invention, the server manages the question contents by natural language processing to extract and manage words of high interest, and provides the user with a plurality of highly relevant questions based on the interest. Since it is a question answering system, it is possible to provide multiple related questions to the user depending on the degree of interest,
There is an effect that the accuracy of the response can be improved.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram showing a question item input screen.

FIG. 3 is an explanatory diagram showing a screen after inputting question items.

FIG. 4 is a flowchart showing the processing of this system.

FIG. 5 is a schematic configuration diagram of an online manual service system according to another embodiment of the present invention.

[Explanation of symbols]

1 ... Server, 2 ... Database (DB), 3 ... Operation terminal, 4 ... Customer PC, 5 ... Maker PC

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Hideyuki Sano             5th valley, 3-38, Motoazabu, Minato-ku, Tokyo             Sawa Building 5th floor, inside Neo Genesis Co., Ltd.

Claims (14)

[Claims] 1. A question answering system comprising a server that receives a question from a user via the Internet and answers the question, and a database that stores the answer to the question, wherein the server receives the question. When a question is entered on the screen, the content of the question is processed in natural language, and in order to confirm the content of the question, the user is asked a question to realize an interactive process. As a result of the interactive process, the question of the user is displayed. A question answering system, which is a server that selects and displays a plurality of highly relevant questions from the database and displays a corresponding answer when any one of the plurality of questions is selected. 2. The server displays a plurality of representative questions on a screen for inputting a question, and when any one of the plurality of questions is selected, the corresponding answer can be displayed.
The question answering system according to claim 1, wherein the question answering system is a server which processes a question content in natural language when the question is inputted. 3. The server, on the screen for inputting a question, selects and displays a plurality of highly relevant questions from the database, which is judged from the previous access log, and displays any one of the plurality of questions. 2. The question answering system according to claim 1, wherein the corresponding answer can be displayed, and the question answering system is a server which processes the question contents in a natural language when the question is input. 4. The question answering system according to claim 3, wherein the server is a server that displays a message regarding a content previously asked from the previous access log on a screen for login or input of a question. 5. A question answering system comprising a server that accepts a question from a user via the Internet and answers the question, and a database that stores the answer to the question, wherein the server receives the question. When a question is entered on the screen, the content of the question is processed in natural language, and in order to confirm the content of the question, the user is asked a question to realize an interactive process. As a result of the interactive process, the question of the user is displayed. A server that displays an answer and selects and displays a plurality of questions that are highly relevant to the user's question from the database, and displays a corresponding answer when any one of the plurality of questions is selected. Question answering system. 6. The server stores the relationship between the word on the question side and the word on the answer side in a database, and when the question is input, extracts the word on the question side by natural language processing and dialogue processing, and Is searched for in the database using the word on the answer side, the answer to the question of the user is displayed, and a plurality of questions that are highly related to the question of the user are selected from the database. 6. The server is a server for displaying as a single display.
Question answering system described. 7. The question answering system according to claim 5, wherein the content of the answer is online manual data, medical information, or real estate information. 8. The question according to claim 1, wherein the server is a server which stores a word relating to a question acquired in the course of the interactive processing in a database and uses it for answering to the subsequent questions. Response system. 9. The question answering system according to claim 1, wherein the server is a server that prompts the message board to leave a question when the user's desired answer is not provided. 10. The server is characterized in that, when a question left on the message board is answered, the server additionally stores the question and the answer in a database and uses them for answering subsequent questions. The question answering system according to claim 9. 11. The question answering system according to claim 10, wherein the server is an electronic bulletin board in which a question is written and an answer is obtained according to a user's instruction, or a server which provides a chat for directly interacting with an operator. . 12. The server is a server for additionally storing a question and an answer corresponding to a question written in an electronic bulletin board or a question and an answer in a chat in a database and using the answer for a subsequent question. Claim 1
The question answering system described in 1. 13. The server according to claim 1, wherein the server is a server that analyzes the words acquired in the course of the interactive processing with the user and the content of the question from the user, and collects information related to the user. 12. The question answering system according to 12. 14. The server is a server which performs natural language processing of question contents to extract and manage words having a high degree of interest, and provides a plurality of questions having high relevance to the user based on the degree of interest. The question answering system according to any one of claims 1 to 3, which is characterized in that.