JPH02236634A - Knowledge base construction support method and device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention provides a knowledge base construction method that efficiently constructs a knowledge base by providing means for supporting knowledge organization and automatically converting it into a knowledge base file. assistive methods and devices;
In particular, if the knowledge has many parts written in Japanese, the amount of knowledge to be handled is large and the knowledge base needs to be modified, and if the knowledge base is modified, other parts of the knowledge base may also be affected. The present invention relates to a knowledge base construction support method and apparatus suitable for various cases.

[Conventional technology]

As a conventional knowledge base construction method, the Information Processing Society of Japan
As described in the Proceedings of the 6th (first half of 1988) National Conference, pages 1483 to 1484, "Japan Univac Co., Ltd., MAKEJ, a knowledge acquisition method for developing expert systems," A plurality of worksheets were provided. [Problem to be solved by the invention] In the above conventional technology, each worksheet is created step by step; Since there is no (electrical) relationship, if a user attempts to partially change the worksheet entries to correct knowledge, the user must manually correct each worksheet related to the change.

■Even after the knowledge has been organized, the user must manually convert it into rules and frame format. ■If the full-width half-width, okurikana, or kana-kanji characters are different when inputting information, even if the input originally has the same meaning, the computer will judge it as different information, which will interfere with inference.

■When a large amount of knowledge becomes available, it may be necessary to create a worksheet that acts as an index to organize it.

■If the user modifies the knowledge base file directly with an editor, etc., the worksheet must be modified one by one later to ensure consistency between the worksheet and the knowledge base. ■It is not easy to understand the effect that modifying one part of the knowledge base will have on other parts of the knowledge base.

There were problems like this.

The purpose of the present invention is to reduce the input operations from the keyboard as much as possible when building a knowledge base, thereby reducing the burden on the user.
A knowledge base construction support method and device that improves the efficiency of knowledge base construction and refinement by ensuring consistency of knowledge between related worksheets and knowledge base files, reducing the effort required to learn the descriptive grammar of knowledge bases, and improving the efficiency of knowledge base construction and refinement. The purpose is to provide. [Means for Solving the Problems] In order to achieve the above object, the present invention first creates a worksheet in the format of a word processor document file, and uses a program that performs input/output to this worksheet. I decided to do so.

Also, since re-entering the same word can cause input errors, we tried to extract the same word one step above the worksheet whenever possible.

Next, parts of the worksheet creation that can be done mechanically are automatically created.

Furthermore, from organizing your knowledge and filling out worksheets,
Automatic conversion to rules and frames that match the grammar of the inference mechanism.

When creating or converting a worksheet, or converting it to a rule or frame, it is now checked for errors such as double definitions. Additionally, it is now possible to dynamically understand the impact that changes in knowledge have on the knowledge base.

Finally, we made it possible to convert knowledge from rules and frames into worksheets.

[Effect]

The character string extraction function in worksheet creation support specifies two points on the document displayed on the display and extracts the character string between them as a word. This eliminates the need for the user to re-enter a character string that has already been entered from the keyboard, thereby eliminating input errors and reducing input effort.

The above string extraction function, when a new string is specified,
Checks the string against the previously specified strings. This makes it possible to detect duplicate extraction of words.

The worksheet creation support function in the worksheet creation support fills in mechanically determined portions on the related worksheet based on the information obtained by the character string extraction function. As a result, the user has fewer input operations from the keyboard, which reduces the possibility of erroneous input and reduces operation time, allowing the user to focus on refining the knowledge base accordingly.

When converting between worksheets, conversion is automatically performed from one to the other between two related worksheets. This eliminates the need for the user to create a conversion destination worksheet himself, which saves time and prevents the contents of the worksheet from being converted by mistake. Furthermore, it is possible to automatically create worksheets that serve as indexes. When converting a knowledge base, the knowledge part related to the knowledge item you want to change is extracted, and changes in inference results caused by changes in knowledge are dynamically displayed, making it easy to understand the impact of knowledge changes on the knowledge base. This allows for appropriate changes and modifications to the knowledge base. When converting rules and frames, information is extracted from the worksheet where knowledge organization has been completed, and after checking the information for inconsistencies, it is converted into rule and frame files. This allows errors such as double definition of frames and references to non-existent frames to be detected, allowing for efficient correction of knowledge. In addition, since it automatically converts rules and frames, users can save the trouble of inputting rules and frames from the keyboard, and at the same time, they are freed from the trouble of learning the syntax for writing rules and frames. and frame files can be created. Furthermore, since knowledge is converted from rules and frames into a worksheet format, users can freely modify the knowledge base file with an editor, etc., without being restricted by tools, and can easily convert the worksheet and knowledge base file later. You can maintain consistency between files. [Example] Hereinafter, as an example of the present invention, (1) an example of constructing a knowledge base of a search-type expert system that searches documents from selected keywords; (2) an example of constructing a knowledge base of a search-type expert system as described above; Among these, degree of association (degree of strength of association between keywords that is added to keywords associated with the selected keyword), degree of association (the degree of association between keywords that is added to keywords associated with the selected keyword), degree of association (the degree of association between keywords that are added to keywords associated with the selected keyword), degree of association (the degree of association between keywords that are added to keywords associated with the selected keyword), and degree of association (the degree of association between keywords that are added to keywords associated with the selected keyword), and degree of association (the degree of association between keywords that are added to keywords associated with the selected keyword). Example of refining the knowledge base by changing the strength of the relationship between keywords and documents (3) Automatically sorting telegrams (such as telex) received by banks, etc. to each business department An example of constructing a knowledge base for a telegram analysis system will be explained with reference to FIGS. 1 to 22.

An overview of the processing in the embodiment will be explained with reference to FIG. 1, FIG. 4, and FIG. 15. In the first step 100 of supporting the task of filling in knowledge on the worksheet, the keyword extraction source worksheet and the search document worksheet 4 are
It supports the extraction of keywords from 30 and the entry of the search document from which the keywords are extracted, keywords, relevance, etc., on the related worksheet and relevance rule definition worksheet 435. In the second step 110 of automatically converting a worksheet into another worksheet with a different method of organizing knowledge, the relationship between the search document worksheet 430 from which keywords are extracted and the keyword is searched from the keyword to the keyword extraction source. document worksheet 430
The knowledge entered in the source worksheet is replaced with the knowledge entered in the source worksheet and output to the destination worksheet. ``A tertiary system that supports the refinement or modification of knowledge.
In step 120, during the knowledge correction work, the test search results are displayed on the association degree/relevance update screen 1500, and when the association degree (relevance degree) is corrected on the screen, the corrected association degree (relevance degree) is displayed. The test search is re-performed based on the , and displayed on the associativity/relevance update screen 1500 again. Furthermore, when the correction work is completed, the corrected knowledge is recorded on the worksheet. ” Fourth step 1 to automatically convert worksheets into knowledge base files
30, a knowledge base file (relevance rule file 445) such as rules and frames is extracted from the worksheet for which knowledge organization has been completed (related document definition worksheet 440).
Perform automatic conversion to . In the fifth step of automatically converting a knowledge base file into a worksheet, conversely, a worksheet for organizing knowledge (related document definition worksheet 440) is converted from a knowledge base file such as rules and frames (relevance rule file 445). Perform automatic conversion to .

Next, an overview of the processing means of the present invention will be explained with reference to FIG. The inference mechanism 210 performs inference based on the constructed knowledge representation (association rules, etc.). The knowledge base storage device 220 stores the constructed knowledge expression files (association rule file 445, etc.). The worksheet storage device 230 stores worksheets for organizing knowledge, such as a search document worksheet 430 and a relevance rule worksheet 435. Knowledge base construction support device 240
supports construction of a knowledge base using the knowledge base file stored in the knowledge base storage device 220 and the worksheet file stored in the worksheet storage device 230 as input and output files.

[Embodiment 1] (1) Overall configuration of the embodiment First, the system configuration of an embodiment of the present invention will be described with reference to FIG. The system of this embodiment includes a central processing unit 300 that executes various processes such as inference and word processing, and stores various knowledge and states, an external storage device 310 that stores knowledge base and worksheet information,
A display device 32o that displays search documents and various messages created with a word processor, an input device W330 that performs input for word processing and responses to various messages, and prints documents created with a word processor and operation results of tools. It consists of a printing device 340.

Next, with reference to FIGS. 3 and 4, related diagrams of tools and worksheets according to an embodiment of the present invention and their functional configuration will be described.

The search document worksheet 430 is a search document entered into the word processor 450.

The worksheet creation support tool 400 displays the contents of the search document worksheet 430 on the display device 320.

The user specifies a character string on the search document as a keyword from the input device 330, the specified keyword is compared with the contents of the keyword information file 405, duplicate specification errors are detected, and the result is output to the display device 320. If there is no error, it is registered in the keyword information file 405 as a new keyword. The worksheet creation support tool 4001t performs entries in related worksheets, association degree rule definitions, and parts of the worksheet 435 that can be entered mechanically. The user uses the word processor 450 to add or modify the degree of association, etc. to the partially filled-in association degree rule definition worksheet 435, thereby completing the association degree rule definition worksheet 435. The inter-worksheet conversion tool 410 converts the correspondence from documents to keywords into the correspondence from keywords to document l\, and outputs it to the related document definition worksheet 440, in order to support the user's knowledge organization. The rule file creation tool 415 acquires information from the related document definition worksheet 440 for which knowledge has been organized, and creates the relevance rule file 4.
Create 450. At this time, the acquired information is compared with information acquired previously and stored in the worksheet information file 420, and errors such as the use of unregistered keywords are detected and displayed on the display device 320. The rule file inverse conversion tool 425 acquires information from the association rule file 445 and fills in the association rule definition worksheet 435.

(2) Operation of the Embodiment The operation of the embodiment of the present invention will be explained using FIGS. 4 to 11. First, the processing of the worksheet creation support tool 400 will be explained with reference to FIGS. 4 to 6 and FIG. 9. Step 900 allows the user to input the file name of the search document worksheet 430 as the input file name, and the file name of the relevance rule definition worksheet 435 to be output as the output file name. In step 910, the contents of the search document 500 are displayed on the display device 320 from the first page. In step 920, the user is asked to specify two points on the document displayed on the display device 320.
Get the string between points as a keyword. Step 9
In step 30, it is checked whether the specified keyword is duplicated with a keyword already specified in the same document, and if it is not duplicated, information such as the keyword's classification name and 5 o sound is stored in step 940. Let the user set it. Step 950 writes the keyword name and its information into the keyword information file 405 and saves it. In step 960, when it is confirmed that the above-mentioned keyword extraction has been completed for all documents, the process proceeds to step 970, and the document is extracted as shown in FIG. Nα600,
Fill in items such as title 610 and keyword 620 to assist in creating a worksheet. Next, the process of the inter-worksheet conversion tool 410 is
This will be explained with reference to FIGS. 6, 7, and 10. In step 1000, when the user inputs the conversion source relevance rule definition worksheet file name and the conversion destination related document definition worksheet file name, the document Ha 600, keyword 620, and relevance 630 are input in step 1010. Obtain the information of the association rule definition worksheet files as a set. In step 102o, the document->keyword relationship obtained in step 1010 is replaced with the keyword->document relationship, and the keyword 700, document Nα 710, and degree of association 720 are written in the related document definition worksheet 440, which has been given a framework in advance.

Next, the process of the rule file creation tool 415 is
This will be explained with reference to FIGS. 7, 8, and 11. In step 1100, when the user inputs the related document definition worksheet file name and the relevance rule file name, in step 1110, the keyword 70
0, Document&710, and Relevance 720 are set as a set, and information on the related document definition worksheet 440 is acquired. Next, in step 1120, a check is made to see if there are any errors in the acquired information, such as duplication or use of undefined frames. If there is an error, the process proceeds to step 1130 and requests correction by the user. If there is no error, the process proceeds to step 1140, and the acquired information is temporarily saved in the worksheet information file 420.

In step 1150, when it is confirmed that all information acquisition has been completed, the process proceeds to step 1160, and the eighth
Output the association rules as shown in the figure. Finally, the processing of the rule file inverse conversion tool 425 will be explained with reference to FIG. Rule file reverse conversion tool 42
5 obtains information regarding the keyword and relevance part from the relevance rule file 445 and fills in the related document definition worksheet 440 and the relevance rule definition worksheet 435. (3) Effects specific to this embodiment According to this embodiment, there is an effect that each keyword can be registered as a keyword without having to be manually input directly from a keyboard. ■An effect is that you can easily specify the keyword classification and Japanese syllabary for the registered keyword.

■An effect is that the keywords extracted from each document are automatically entered into the worksheet. ■Each keyword is
This has the effect of automatically creating a list showing which documents are associated with what weight, which helps in setting appropriate weights.

■Completing worksheets and rules that do not require thinking. The creation of the aisles is done entirely by the tool, so the user does not have to worry about it and has the advantage of being able to focus on refining their knowledge (in this case, setting the degree of relevance).

■Since the tool automatically converts the rules, there is no need to be familiar with the writing syntax of the rules. ■Since the reverse conversion from rules to worksheets is performed, you can freely modify the knowledge base file with an editor, etc., without being bound by the rules, and easily create a worksheet that matches the contents of the knowledge base file later. [Embodiment 2] (1) Overall configuration of the embodiment First, the system configuration of an embodiment of the present invention will be explained with reference to FIG. The system of this embodiment includes a central processing unit 300 that reads knowledge written on a worksheet, calculates degrees of association and relevance, executes various processes, and stores various types of knowledge and states; An external storage device 310 that stores information on bases and worksheets, a display device 320 that displays selected keywords, searched document names, various messages, etc. It consists of an input device 330 for inputting data, and a printing device 340 for printing worksheets and the like output as a result of the operation of the tool. Next, with reference to FIG. 3, FIG. 12, FIG. 14, and FIG. 15, the related diagrams of tools and worksheets according to an embodiment of the present invention and their functional configuration will be explained.

The search document definition worksheet 1-1210 defines the document number, title, etc. of the document to be searched. The alphabetical keyword list worksheet 1220 is a worksheet in which registered keywords are arranged in alphabetical order. The keyword list by category worksheet 1230 organizes registered keywords by keyword category. Associative keyword definition worksheet 1240
defines the associative relationship and degree of association of keywords. The related document definition worksheet 1250 defines the relationship and degree of relevance between keywords and search documents. The degree of association/relevance update tool 1200 includes a search document definition worksheet 1210, a list of 50 syllabary keywords worksheet 1220, and a list of keywords by classification worksheet 12.
30, associative keyword definition worksheet 1240, and related document definition worksheet 1250, and read the contents of search document definition worksheet 1210, alphabetical keyword list worksheet 1220, classification keyword list worksheet 1230, and associative keywords. If there is no contradiction between the contents of the definition worksheet 1240 and the related document definition worksheet 1250, the association source keyword input screen 1400 is displayed on the display device 320, and the user is prompted to input the association source keyword from the input device 330. From the association source keyword, the currently set keyword association relationship and degree of association, the relationship and degree of association between the keyword and the searched document, and the document to be finally searched and the ranking and degree of the search are output to the display device 320. ．． The degree of association and degree of association to be updated are input by the user through the input device 1330, and the degree of final search is recalculated.

When all updates are completed, the latest knowledge is output to the associative keyword definition worksheet (updated) 1260 and the related document definition worksheet (updated) 1270. (2) Operation of the embodiment Regarding the operation of the embodiment of the present invention, FIGS.
This will be explained using Figure 18. The processing of the association degree/relevance degree update tool 1200 will be explained with reference to FIGS. 12 and 13. In step 13oO, as the input file name,
Search document definition worksheet 1210, alphabetical keyword list worksheet 1220, classification keyword list worksheet 1230, associative keyword definition worksheet 1240. and the related document definition worksheet 1250, respectively.

In step 1310, the search document definition worksheet 12
10, alphabetical keyword list worksheet 1220,
and the contents of the classified keyword list worksheet 1230, and create an internal data structure as shown in Figure 16. After that, the contents of the associative keyword definition worksheet 1240 and the related document definition worksheet 1250 are read and registered in the search document definition worksheet 1210, the alphabetical keyword list worksheet 1220, and the classified keyword list worksheet 1230. Check that there are no contradictions with the keywords and document content, and
The information is stored in a related document table as shown in Figure 7 and an associated keyword table as shown in Figure 18. In step 1320, the registered keywords are displayed on the association source keyword specification screen 140o. Step 134
0 allows the user to specify the association source keyword. In step 1350, the associative keyword table of FIG. 17 and the related document table of FIG. 18 are referred to from the association source keyword, and the currently set keyword association relationship and degree of association, the relationship and degree of association between keyword documents, The documents to be finally retrieved and their degrees are searched and calculated, and displayed in the association degree/relevance update image 150o. In step 1360, the user is asked to set the degree of association and degree of association to be updated. In step 1370, the degree of retrieval of the document is recalculated based on the updated association degree/relevance degree and displayed on a new association degree/relevance screen. In step 1380, the degree of association and degree of association are updated and created until the update results are satisfied.

In step 1330, when it is selected to finish the update work, the process moves to step 1390, and the latest contents (keyword associations, relationships between keyword search documents) are transferred to an associative keyword definition worksheet 1240 with a framework given in advance.
and output to the related document definition worksheet 1250.

(3) Effects specific to the embodiment According to this embodiment, ■Associativity and relevance can be updated continuously until the search results are satisfied, so convergence to ideal associativity and relevance is quick. effective.

■The document search results are updated on the spot by updating the degree of association and relevance, which has the effect of eliminating the need to create an executable form of the expert system each time the knowledge base is updated. ■By updating the degree of association and degree of association, the document search results are updated on the spot, which has the effect that setting the degree of association and degree of association through trial and error can be completed in a shorter time.

■Since the final updated results of the degree of association and degree of association are automatically output to the association definition worksheet and related document definition worksheet, the user can save the effort of frequently updating the worksheet.

[Embodiment 3] (1) Overall configuration of the embodiment First, the system configuration to which the present invention is applied will be explained with reference to FIG. The system to which the present invention is applied includes a telegram receiving device 1900 that receives telegrams from the outside, reads telegrams and business specifications, reads knowledge written in worksheets, and performs various processes such as calculating relevance. Central processing unit 3 that executes and stores various knowledge, status, etc.
00, external storage device 310 that stores knowledge base and worksheet information, display device 13 that displays telegrams, business specifications, set keywords, various messages, etc.
0, an input device 1!330 for setting keywords and inputting updated values of relevance, and a printing device 340 for printing worksheets and the like outputted as the operation results of the tool.

Next, with reference to FIG. 14 and FIGS. 19 to 22, related diagrams of tools and worksheets according to an embodiment of the present invention and their functional configurations will be described.

The telegram file 2000 is a telegram received by the telegraph receiving device 19oO that has been changed into a file format that can be accessed by the worksheet creation support tool 400. The work specification file 2010 is a document that summarizes the work contents of each business department such as a bank. The related department definition worksheet 2020 defines the relationship between keywords extracted from telegrams and business specifications and each business department, and the degree of relationship. The worksheet creation support tool 400 reads a telegram or business specification, displays it on the display device 320, and allows the user to specify a word as a keyword using the input device 330.

Further, the user is allowed to set the relationship between the keyword specified by the user and each business department, and the strength of the relationship (degree of association).

The relevance update tool 1200 reads the contents of the related department definition worksheet 2020 and displays the keyword input screen 14.
00 is displayed on the display device, and the user is prompted to input a keyword from the input device 330. Based on the input keyword, the relationship and degree of association between the keyword and each business department, as well as the finally designated business department and the ranking and degree of designation are output to the display device 320. The user manually inputs the degree of association to be updated using the input device 330, and finally the specified business department and its degree are recalculated. When all updates are completed 2120L, the latest knowledge is output to the related department definition worksheet (updated) 2030. The rule file creation tool 415 converts the related department definition worksheet 2030 into a rule format description file 2o4.
Automatically converted to 0.

(2) Operation of the Embodiment The operation of the embodiment of the present invention will be explained using FIG. 14 and FIGS. 19 to 22.

First, the process of the worksheet creation support tool 400 is
This will be explained with reference to FIG. 4 and FIGS. 19 to 21. Step 2100 prompts the user to input the file name of the telegram file 2000 or the business specification file 2010 as the input file name. Step 2
At step 110, the telegram file 2000 or business specification file 2010 is read and displayed on the display device 320. In step 2130, the user is asked to specify a keyword and a character string to be set. In step 2140, the user is allowed to set the business department name and the degree of association (degree of association) related to the keyword specified by the user.

All keyword settings and associative relationship settings are complete 2
120, the process proceeds to step 2150 and outputs the related department definition worksheet 202o. Next, the processing of the relevance update tool 1200 will be explained with reference to FIGS. 14, 19, 20, and 22. In step 2200, the user is prompted to input the file name of the related department definition worksheet 2o30 as the input file name.

In step 2210, the related department definition worksheet 20
Read the contents of 30 and create a related department table.

Furthermore, the registered keywords are displayed on the keyword specification screen 1400. In step 2230, the user is prompted to specify a keyword using the input device 330. In step 2240, the related department table is referred to based on the input keyword, and the relationship and degree of relationship between the currently set keyword and business department, as well as the department to be finally specified and its degree, are searched and calculated. and displayed on the relevance update screen l500. In step 2250, the user is asked to set the degree of association to be updated. In step 226o, the degree to which a business department is specified is recalculated based on the updated degree of association and displayed on the degree of association update screen l500. In step 2270, this update operation is repeated. If the end of the update work is selected in step 2220, the process moves to step 2280, and the latest contents (keywords, relationships between business departments) are output to the related department definition worksheet 2030, which has a framework given in advance.

[Effects specific to the embodiment]

According to this embodiment, there is an effect that it can support the construction of knowledge bases for rule-based systems other than expert systems. ■Keywords can be extracted from the original text of the telegram or business specifications, but if the tool is in an accessible file format, it can be used in either case. ■If one keyword is related to multiple business departments, there are two ways to identify the department: one is to identify the department based on the degree of relevance from one keyword, and the other is to identify the department differentially from multiple keywords. Although it is conceivable, it has the effect of being able to handle both cases. [Effects of the Invention] According to the present invention, first of all, the number of operations from the keyboard can be reduced as much as possible, thereby shortening the time spent building the knowledge base of the expert system and reducing the burden on the user. At the same time, it has the effect of reducing operational errors and improving the quality and efficiency of knowledge base construction. Second, since you can use word processors and editors to create and modify worksheets and knowledge base files, you can easily make small modifications to worksheets and knowledge base files using the word processors and editors you are familiar with. There is. Third, since worksheets are automatically converted into knowledge base files, a knowledge base can be constructed by someone who does not know rules or frames (without being aware of knowledge description grammar).

Fourth, since the knowledge base file is automatically converted to a worksheet, even if the user directly modifies the knowledge base file, it is easy to maintain consistency between the worksheet and the knowledge base afterwards. There is an effect that it can be done.

Fifth, since the contents of the worksheet and the knowledge base can be easily matched, the knowledge base can be maintained and refined by using the worksheet, which is easier to understand.

Sixth, since each step (means) of knowledge base construction support is independent using the worksheet as a medium, it is possible to perform the necessary parts (only the necessary parts) of the series of knowledge base construction tasks. It is possible to construct a knowledge base without being bound by knowledge base construction support means.

Seventh, since confidence determination guidance is provided, confidence can be made consistent within the knowledge base.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an overview of knowledge base construction support processing, FIG. 2 is a diagram showing an overview of the overall configuration including knowledge base construction support, and FIG. 3 is a diagram showing the first and second implementations of the present invention. FIG. 4 is a diagram showing the relationship between tools and worksheets in the first embodiment of the present invention, and FIG. 5 is a diagram showing the functional configuration of the example. Figure 6 shows a worksheet from which keywords are extracted, and Figure 6 is a worksheet that summarizes the degree of association from documents to keywords in the first embodiment of the present invention. In the example, a worksheet summarizing the degree of relevance from keywords to documents, FIG. 8 is a rule file of the first embodiment of the present invention, and FIG. Figure 10 is a flowchart showing the processing procedure of the inter-worksheet conversion tool.
FIG. 1 is a flowchart showing the processing procedure of the rule file creation tool, FIG. 12 is a diagram showing the relationship between the tool and worksheet of the second embodiment of the present invention, and FIG. 13 is a flow chart showing the processing procedure of the rule file creation tool.
A flowchart showing the processing procedure of the associativity/relevance update tool, Fig. 14 is a diagram showing the keyword specification screen of the associativity/relevance update tool, and Fig. 15 shows the associativity/relevance update tool of the associativity/relevance update tool. Diagram showing the relevance update screen, 16th
17 is a diagram showing an internal data structure regarding documents and keywords, FIG. 17 is a diagram showing a table regarding the degree of association, FIG. 18 is a diagram showing a table regarding the degree of association, and FIG. 19 is a diagram showing the third aspect of the present invention. Diagram showing the functional configuration of the embodiment, No. 20
21 is a flowchart showing the processing procedure of the worksheet creation support tool, and FIG.
3 is a flowchart showing a processing procedure of a relevance update tool. Figure 1-3 Figure 4 Figure 13 Figure 14 Figure 13 Figure, Figure 16 Half 19 Figure 20 Patterns

Claims (1)

[Scope of Claims] 1. A worksheet is provided that defines a method for organizing knowledge, and (a) a first step of supporting the entry of knowledge into the worksheet; (b) the worksheet is used as a knowledge (c) A third step of supporting the refinement or modification of knowledge; (d) Automatically converting the worksheet into a knowledge base file. A knowledge base construction support method comprising: a fourth step; and (c) a fifth step of automatically converting the knowledge base file into a worksheet. 2. The first step above is to obtain the knowledge information of the worksheet, extract the string from the document, define the information to be added to the string, define the information to add the string information, and extract the string from the document. It is characterized by adding information, associating character strings, setting confidence level, and outputting it to a worksheet.
A knowledge base construction support method according to claim 1. 3. The knowledge base construction support method according to claim 1, wherein the second step reads the knowledge information on the worksheet, reorganizes the knowledge, and outputs it to another worksheet. 4. The third step above is to read the knowledge information on the worksheet, input the knowledge change item, extract the knowledge part related to the change, and input the knowledge change content to display the change in the inference result due to the change. 2. The knowledge base construction support method according to claim 1, wherein the knowledge base construction support method according to claim 1 is characterized in that the knowledge portions related to the changes are corrected and the knowledge is output on a worksheet. 5. The knowledge base construction support method according to claim 1, wherein the fourth step reads the knowledge information on the worksheet, organizes and stores the knowledge, and outputs a knowledge base file. . 6. The knowledge base construction support method according to claim 1, wherein the fifth step reads the knowledge information in the knowledge base file, organizes and stores the knowledge, and outputs a worksheet. . 7. The knowledge base construction support method according to claim 1, wherein in each of the first to fifth steps, the content of knowledge information read from a worksheet or a knowledge base file is checked. 8. The output worksheets and knowledge base files are
7. A method for assisting experts in building a knowledge base according to any one of claims 2 to 6, characterized in that the method can be accessed using a word processor or an editor. 9. The third step is characterized in that it is possible to continuously input changes in knowledge and dynamically check changes in inference results due to changes in knowledge.
Knowledge base construction support method described in section. 10. A knowledge base construction support device comprising an inference means, a knowledge base storage means, a worksheet storage means, and a knowledge base construction support means. 11. The knowledge base construction support means includes a worksheet creation support means that supports interactive creation of worksheets, an inter-worksheet conversion means that automatically creates related worksheets from worksheets, and a knowledge refinement or modification work. The present invention is characterized by comprising a supporting knowledge base update means, a knowledge base file conversion means for converting a worksheet into a knowledge base file, and a worksheet inverse conversion means for inversely converting the knowledge base file into a worksheet. A knowledge base construction support device according to claim 10. 12. A patent claim characterized in that the worksheet creation support means, the inter-worksheet conversion means, the knowledge base file conversion means, and the worksheet inverse conversion means are each independent using the worksheet as a medium. Range 1
The knowledge base construction support device described in item 0.