JPH0415828A - Method and system for inference - 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] [Industrial Application Field] The present invention relates to a method and system for making inferences using a knowledge base, and in particular, even a user without specialized knowledge can make inquiries from the system. The present invention relates to an inference method and device that can accurately answer questions.

[Prior Art] In general, the following methods are known as information inquiry methods for inference systems using conventional rule-based knowledge.

(1) Information inquiry method during forward inference Extract the rule to be evaluated from the knowledge base, determine which conditional clauses are written in the condition section of the extracted rule, and determine which information is undetermined. The expression of the rule description is presented on the display device and input of its truth value and confidence level is accepted.

(2) How to query information during backward reasoning Take out a rule that has the information that is the goal hypothesis in the rule conclusion part, use the information described in the condition part of this rule as the goal hypothesis, and create a rule that has this in the rule conclusion part. By searching and chaining rules one after another, we finally determine unconfirmed information written in the rule condition section that is not chained, and display its truth value and certainty as a display device using the expression of the rule description. It accepts input of degrees.

In addition, as a rule description format, relationships between information are expressed as IF-
It is expressed in the framework of THEN format, and does not include procedures to be activated when an inquiry is made.

A known example of this is ES/KERNEL's backward inference. This backward reasoning uses information called private notes written in working memory called view notes, and the reasoning mode is set to 1.
When "interactive" is specified, an inquiry is made to the user regarding undeterminable event information using the private memo expression described in the rule.

Regarding the information inquiry method using backward reasoning, for example, see "Hitachi Creative Workstation 2050 ES/KERNEL/W Explanation (Manual) Manual No. 2050-3-6234, (No. 35)
Pages 37 to 37).

[Problems to be Solved by the Invention] In the above-mentioned prior art, the query sentence is displayed as it is written in the knowledge base during inference.

Therefore, an inquiry using technical terminology will be presented. In this case, the system user can
If you are an amateur in the business of making judgments and diagnoses based on specialized knowledge, such as diagnosing the failure of various types of equipment, a situation may arise in which it is difficult to fully understand the meaning of the inquiry text.

Furthermore, in order for the user to accurately answer the inquiries presented by the system, it will be necessary to refer to related materials such as necessary databases and past reports. For example, to determine the event that "sales are on an upward trend", it is sufficient to refer to trends in sales data.

However, the interpretation of ``on the rise'' differs from person to person, and it cannot be precisely defined, so in the conventional method, this ``sales are on the rise'' is the terminal event in relation to the knowledge base data, and the reference There was a problem in that it was not possible to describe the relationship between what information should be used.

Conventionally, as one of the guidance functions, when a user who does not know how to use a program makes an inquiry to the system by issuing a command called "rHELP", the system does not know how to use the program. However, this is only activated by a command issued by the user, and is accompanied by an inquiry issued by the system that automatically provides an explanation that allows the user to accurately understand the meaning of the inquiry. There is no known system that outputs this.

As shown above, in all conventional methods, when the user does not understand the meaning of a query statement well when making inferences using a knowledge base, or when the user is unable to properly answer the question There was a problem in that the system did not give sufficient consideration to countermeasures for situations where the user did not know what to do, and it was not efficient for the user to respond after fully understanding the inquiry text from the system. .

Therefore, an object of the present invention is to provide a method and a device for explaining technical terms when a query includes such terms in reasoning using a knowledge base.

Another object of the present invention is to provide a method and apparatus for extracting appropriate materials and data to be referred to in order to accurately answer a query when making inferences using a knowledge base. be.

[Means for Solving the Problems] In order to achieve the above object, the inference method and device of the present invention as a basic concept include predetermined information in the query text that is output from the computer system. Automatically detect (technical terminology) and use this information to provide auxiliary data (explanation, etc.) about this information (technical terminology).
It is characterized by being configured to output both.

Specifically, the present invention stores, on the storage device, assistance related to the field targeted by the system (for example, the business type that the system processes, such as bank loan appraisal work, automobile failure diagnosis work, etc.). Data (corresponding explanations such as meanings and examples of technical terms used in the field) are stored in a knowledge base so that they can be searched using predetermined information (technical terms). output the query text that has been specified to the output device, determine whether or not the output query text contains the predetermined information, and as a result of this determination, if the predetermined information is included, the The auxiliary data stored in the storage device is searched using the predetermined information in the output query sentence, the retrieved auxiliary data is output to the output device, and the output query sentence and the Information corresponding to the output auxiliary data (answers to inquiries, information for erasing the display of the auxiliary data, display instructions for the next page, etc.) is input from the input device.

Further, after outputting the inquiry text, inquiry information regarding whether or not to search the auxiliary data stored in the storage device using the predetermined information is output to the output device, and a response to this inquiry information is output. Information is input from the input device, and when the input response information means that the auxiliary data is to be searched, it is determined whether the predetermined information is included in the output query text. It is.

Furthermore, in order to achieve the other objectives mentioned above, in addition to the query statement, programs related to this query statement are included in the information representing the knowledge stored in the knowledge base (the rule IF part of the IF-THEN format). Include a name (for example, a program that displays a graph or outputs a form), stores a program corresponding to this program name in a storage device, and outputs information representing the knowledge stored in the knowledge base. When the information representing the outputted knowledge includes a program name, a command for executing the program stored in the storage device corresponding to the program name is input from the input device. , in response to the input command, executes the program and outputs the execution result to the output device.

[Effect] The operation based on the above configuration will be explained.

According to the present invention, by storing auxiliary data (explanation) related to the field targeted by the system on the storage device so as to be searchable using predetermined information (technical terminology), the field targeted by the system is Information such as detailed explanations and examples of terms can be retrieved by providing the terms, so even users who are not experts can accurately understand the content of the inquiry and respond correctly.

Furthermore, the query text stored in the knowledge base is output to the output device, it is determined whether or not the predetermined information is included in the output query text, and as a result of this determination, it is determined whether or not the predetermined information is included. If so, the predetermined information in the output query text is used to search the data stored in the storage device, and the searched data is output to the output device to process the query. Explanatory and illustrative information corresponding to terms in a sentence can be displayed.

By inputting information corresponding to the output query text and the output data from the input device, the system user can refer to detailed explanations and example information corresponding to the terms in one query text and make inquiries. It is possible to provide detailed explanations of technical terms in one query to system users who do not fully understand technical terms.

Further, after outputting the inquiry text, inquiry information regarding whether or not to search the data stored in the storage device using the predetermined information is output to the output device, and response information to this inquiry information is output. is input from the input device, and when the input response information means to search the data, the term is determined by determining whether the output query contains the predetermined information. Only when the user determines that an explanation is necessary can a detailed explanation be given as necessary.

In addition, by including the program name in the information representing knowledge stored in the knowledge base, it is possible to associate the name of a specific program that outputs the necessary information with specific knowledge information. . Then, a program corresponding to this program name is stored in a storage device, information representing the knowledge stored in the knowledge base is outputted to an output device, and the program name is included in the information representing the outputted knowledge. inputting a command for executing the program stored in the storage device number corresponding to the program name from the input device, and executing the program in response to the input command; By outputting this execution result to the output device, the user can instruct the presentation of relevant information necessary to accurately answer the knowledge output in the form of a query.

[Example] An embodiment of the present invention will be described in detail below.

FIG. 1 is a flowchart showing a first embodiment of the processing procedure of the present invention, and FIG. 2 is an embodiment of the hardware configuration of a computer device implementing the present invention. Also, Figure 3,
FIGS. 4 and 5 show the configuration of various data stored on the storage device on the first actual flow side.

In FIG. 2, reference numeral 201 is a bitmap display for displaying queries and inference results from the system. A keyboard 202 is a device through which a user inputs input data into the system. A mouse 203 is an input device for pointing on the display 201 to provide an answer menu to a query displayed by the system. The storage device 205 stores a terminology dictionary 20 in which terms related to the target field of the system and their explanations are recorded.
6. Rule-type knowledge memory 207 that stores rules expressed in IF-THEN format; working memory 20 that stores system states such as intermediate hypotheses for inference and input information;
8. It is composed of a cut-height term storage memory 209 that stores terms cut out by analyzing query sentences, and a query sentence storage area 210 that stores query sentences. The processing device 204 is in charge of retrieving, inspecting, and outputting data from each of these devices and various storage areas of the storage device 205, and controls the operation of these devices, and controls the operation of each of the devices according to a program stored therein. The program within this processor also includes a reasoning mechanism. In the above configuration, the term dictionary 205 and the storage memory 2 for cut-out terms
09 is a portion provided as a feature of the present invention.

The table 300 shown in FIG. 3 is the structure of the term dictionary 206 in FIG. 2, and consists of term names 301 of terms used in the target field of the system and explanation contents 302 of the corresponding terms.

FIG. 4 shows the IF stored in the rule type knowledge memory 207.
- This is an example of a THEN format rule, and here, as an example of rules, five simple rules for determining whether it is okay to lend are listed (401, 402,
403, 404, 405).

FIG. 5 shows the table structure of the cut-out term storage memory 209, which is composed of an area 501 that stores a plurality of terms cut out by analyzing the query issued by the system.

FIG. 6 is an example of the screen output of this system, and shows an area 601 for displaying inquiries. Answer menu display areas 602 and 603 for this inquiry. It is composed of a term explanation content output area 604.

Next, the operation of each part shown in FIG. 2 will be explained based on the flowchart shown in FIG.

This process starts when an inquiry is generated by inference execution. The operation process of the inference mechanism up to the generation of an inquiry is performed as shown below.

In this embodiment, a backward inference is activated as an inference method, and the objective is to verify whether or not the event "it is OK to lend" is established. The reasoning mechanism is
Rule 401 stored in rule type knowledge memory 207.
Inference is made by linking 402, 403, 404, and 405. In other words, it is checked whether the target event ``It is OK to make a loan'' is written in the working memory 208. Since ``It is OK to make a loan'' does not exist in the working memory 208, the rule type Among the rules stored in the knowledge memory 207, the rule that has the THEN section written as 1' may be loaned is searched for.Here, the rules 404 and 405 match, so the inference mechanism uses the IF section of this rule. We will further verify whether the conditions described in rule 4 hold true.For example, rule 4
Using 04, in order for ``it is okay to lend'' to hold true, the conditions that ``the lending company is a company that can be trusted in the world'' and the conditions that ``the lending company is doing well in terms of stock'' are satisfied. do it. Then, it is further checked whether these conditions exist in the working memory 208, and if they do not exist, a rule in which these conditions are written in the THEN section is further found from the rule type knowledge memory 207. The function of the inference mechanism is to follow the chain of rules one after another to verify whether the target event ``it is okay to lend'' holds true. This is incorporated as a basic function of the knowledge base system, and is not particularly shown in the flowchart here. When the rules are chained using this inference mechanism, a conditional event is finally reached that does not exist in the working memory 207 and cannot be followed any further in the chain of rules. In this case, the system user is required to interactively input whether the conditional event is satisfied. This is the occurrence of an inquiry, and the conditional event is displayed on the display 201 as an inquiry.

Inference execution by the inference mechanism ends when inquiries regarding all rule chains are completed or when a target event is established.

When an inquiry occurs, this inquiry is displayed in the inquiry display area 601 of the screen on the display 201.
Output to. Also, at this time, memorize the inquiry text! 2
05 is stored in the query storage area 210 (step 101).

After the question text is stored in the inquiry text storage area 210, this question text is divided into various nouns and phrases using existing methods such as natural language analysis and lexical analysis, which are then included in the query text. The term is cut out and stored in the cut term storage memory 209.

For example, if the inquiry text is "The loan recipient is a listed company," as shown in the cut-out term storage memory 209 in FIG. Two terms will be extracted. Here, this term extraction process does not require the use of the lexical analysis method described above; for example, it can be done by extracting only continuous strings of kanji or katakana from the query text. This can also be achieved in a simple way. It is checked whether this cut-out term exists in the term dictionary 206 (step 102).

This checking process is performed as follows.

First, each of the specific cut-out terms 501 from the cut-out term storage memory 209 is retrieved, and whether there is a word whose name matches the character string stored in the term name 301 column of the term dictionary 206 or not. If there is a match, it is said to exist in the term dictionary, and if there is no match, it is assumed that it does not exist.

In this procedure, it is determined whether or not the cut-out term 501 exists in the term dictionary 206, and if it does not exist in the term dictionary 206, the term input by the user using the mouse 203 is determined.
Receive answers to inquiries (step 105)
), this input method is performed by using the mouse 203 to pick any of the answer menu display areas 602 and 603 displayed on the screen on the display 201.

When the answer to the inquiry question from the user is input, the inquiry question and its answer result information are written to the working memory and saved as final information according to the input answer, one process is finished, and the inference is started again. Control is returned to the execution of the function.

On the other hand, as a result of the determination in step 102, the cutout term 501
exists in the term dictionary 206, the cut-out term 5
Step 103) of searching and retrieving the information written in the explanation content 302 column in the terminology dictionary 206 corresponding to 01)
, output the explanation content to the terminology explanation output area 604 on the screen of the display 201 (step 104), and then proceed to step 105. Zero or more steps 101 to 1
05 is repeated each time there is another inquiry.

Here, the procedure for searching and extracting the explanation content corresponding to the cut-out term 501 from the term dictionary 206 is performed in the following manner.

Using the cut-out term 501 as a key, it is checked whether it matches the term name 301, and when there is a match, the explanation content 302 is extracted. This is done for all of the terms 501 cut out in step 102.

According to the embodiment described above, when the system presents an inquiry, if a technical term exists in the inquiry, a detailed explanation of that term is automatically presented, so that the user can This has the effect that the meaning of the inquiry text can be easily and accurately understood. Therefore, the user can easily respond to the terminology explanation and respond to the inquiry in the next step 105. Moreover, only relevant information is presented, so there is no confusion of information.
The user only has to consider what is displayed on the screen. Therefore, even users who are not skilled experts can use the system to improve their understanding of technical terms and the like.

5 In this embodiment, a character string sentence is simply used as an example of the explanation content 302 of the term dictionary, but the invention is not limited to this, and by storing image data in the explanation content 302 column, the term Image data can also be presented as an explanation.

Furthermore, this embodiment can be modified to provide various modifications. This modified example will be explained below.

FIG. 7 is a flowchart showing the processing procedure of a modified embodiment (second embodiment). FIG. 8 shows an example of screen output displayed on the display 201 of the modified embodiment. Here the seventh
The processing in the flowchart shown in the figure is a modification of the processing from step 102 onward in the flowchart shown in FIG. 1, so this modified portion will be explained based on the flowchart.

If it is determined in step 102 that the term 501 extracted from the query exists in the term dictionary 206, the message ``Term explanation available'' 802 is output on the screen of the display 201 as shown in FIG. Display the menu 801 for requesting explanation display (step 701
).

After that, an answer to the explanation display request or inquiry text input by the user using the mouse 203 is accepted (step 702), and the input type (whether to request an explanation or not) is accepted.
Step 703), if the received input is an explanation display request, the term dictionary 20 corresponding to the cut-out term 501 is
step 103) of extracting the explanation content 302 in 6);
This is output to the term explanation output area 604 of the display 201 (step 104). After that, step 70
The process returns to step 2 and enters the input waiting state again.

On the other hand, if it is determined in step 703 that an answer to the inquiry has been input, the process proceeds to step 105, and the inquiry and its answer are written into the working memory as confirmed event information (step 105).
Thereafter, this process is ended and control is returned to the inference mechanism to continue inference.

In this modified embodiment, explanations of technical terms can be presented or not presented according to the user's needs, so if the user is an expert, for example, the system can be used without specifically displaying the explanation contents. You can proceed with the implementation.

In addition, in FIG. 7, step 102 is replaced by step 1.
It may be moved from between 02 and 701 to between steps 703 and 103. In this case, only when the user requires an explanation of technical terms, step 102 (predetermined in the query) It is only necessary to check whether the term is included), so unnecessary checks (step 102 is also unnecessary when no explanation is required) can be omitted.

Furthermore, although this modified embodiment and the above-mentioned embodiments display the explanation of terms in an area separate from the display of the inquiry text, it is also easy to directly replace the terms in the inquiry text with the explanation contents and display them. Can be implemented. As a result, it is possible to display the inquiry text in terms that are easy to understand even for a layman, replacing the content of the inquiry text expressed using technical terms.

Further, in the above-mentioned embodiment, the term explanation content 302 is stored in the storage device 205 as a separate structure from the rule-based knowledge memory 207 that is the knowledge base, but even if this implementation method is not used, for example, Ru-ru type knowledge memory 2
This can also be realized by changing the expression format of the knowledge stored in 07 as shown in FIG. Rule 402 in the same figure is IF
-In addition to the normal information relationships expressed in the THEN format, each event in the rule 11 part can be described in association with the content 302 of the explanation of the term 301 used therein. By doing this, when outputting the inquiry text to the display 201, the explanation 302 of the corresponding term is also output, thereby realizing the same processing as in the above-described embodiment. In this case, 301,302
The rule body can be made easier to see by flickering the part or displaying it in a different font size or color from the main body part of the rule 402.

Next, another embodiment (third embodiment) of the present invention will be described. FIG. 10 is a flowchart showing another embodiment of the present invention. FIG. 11 is an example of a hardware configuration for realizing another embodiment. FIG. 12 is an example of the expression format of the rules stored in the rule type knowledge memory 207 in FIG. 11.

FIG. 13 shows the program storage area 1002 in FIG.
This shows the table structure of . FIG. 14 shows an example of screen output in another embodiment.

The configuration in FIG. 11 is a storage device with the hardware configuration shown in FIG. 2! This table replaces the various tables stored in 205. In this embodiment, when a question is asked and it is difficult to explain it in words as shown in FIGS. 3 and 9, a program is started instead of using words. As shown in FIG. 11, the storage device 205 has IF-
It consists of a rule type knowledge memory 207 that stores THEN format rules, a working memory 208 that stores information that maintains the state of the system, and a program storage area 1002 that stores various additional programs.

Here, IF-T written in the rule-based knowledge memory 207
As shown in FIG. 12, the HEN format rule consists of a precondition event 1102 written in the IF part and a conclusion event 1104 written in the THEN part, which is a rule drawn as a conclusion when this condition is satisfied. In addition to the rule description format shown in the figure, primary information is added.

A program name 1105 can be written in correspondence with each prerequisite event written in the rule IF section. For example, in FIG. 12, the program name ``graph output (ordinary income trend)'' (1105) is written in correspondence with the conditional event ``the ordinary income of the lending company has increased in the past two years'' (1102).

The interpretation of this statement is ``The ordinary income of the lending company is
In order to determine whether or not the situation of ``older age limit'' is established, a graph of trends in ordinary profits can be used as a reference.''

The program that outputs this ordinary profit transition graph is "graph output (ordinary profit transition)" (1105). In addition, the program ``Graph Output (Changes in Capital Investment Costs)'' corresponds to the question ``Capital investment costs of loanee companies have not decreased.'' In this way, for each event that is known in advance, the name of the reference information presentation program used to determine the event can be written. The actual state of this reference information presentation program is in the program storage area 100.
2, and as shown in FIG. 13, it consists of a program name 1201 and 12o2 in which the actual program code is stored.

Next, the processing procedure of another embodiment will be explained according to the flowchart of FIG.

First, this process is started when an inquiry is generated by the inference execution of the knowledge base system.

First, a query is output to the query output area 601 on the screen of the display 201 (step 901).
).

Also, it is checked whether the output query statement 11o2 has the reference information presentation program 1105 in the rule description (step 902), and if it has it, the reference information presentation program 1105, which is the program name, is checked (step 902).
(step 903), and according to the stored name (using this program name as a key), the code of the corresponding program is selected from the table stored in the program storage area 1002. The execution module 1202 is extracted, loaded into the processing device 11204, and executed (step 904).
). As a result, for example, when the system presents an inquiry question such as ``The ordinary profit of a company that receives loans has increased over the past two years'', it will automatically start the program ``graph output (ordinary income trend)'' and display the execution results. display 2
Display on 01. FIG. 14 shows an example of screen output displayed on the display 201 during execution. Screen 1301
The graph shown in is displayed by executing the program "Graph Output (Ordinary Profit Trend)".

Thereafter, the type of the answer menu display area 602 or 603 entered using the mouse 203 is identified for the answer to the inquiry (step 905), and the inquiry and the answer are written into the working memory 208 and stored as final information (step 905). 906). After that, the inference function is executed again.

On the other hand, in the determination at step 902, query statement 11
If the program name 1105 is not written in 02, steps 903 and 904 are skipped and the process proceeds to step 905.

According to this embodiment, a program can be written in advance that causes each of the prerequisite events described in the IF part of an IF-THEN format rule to present information to be referred to when determining that event, and the inference can be executed. When an inquiry occurs, the program is automatically executed and relevant information is presented to the user, allowing the user to refer to the presented information and determine an answer to the inquiry accurately and efficiently. Therefore, the time and effort required to search for related materials among documents and files and to reprocess necessary data can be reduced.

Furthermore, in this embodiment, when an inquiry occurs, the related program is automatically executed, but by modifying this embodiment, the program can be executed only when the user needs it. It is also possible to do this.

The following modified embodiment (fourth embodiment) will be explained based on the flowchart of FIG. 15 and the output display screen of FIG. 16.

The flowchart in FIG. 15 includes step 140 in the process after step 903 in the flowchart shown in FIG.
1,1402.1403 was added.

First, if the prerequisite event displayed as an inquiry question includes a program name in the rule expression (step 9
02), the corresponding program name 1105 is stored (step 903).

Thereafter, a reference information output request menu 1501 is displayed on the display 201 so as to indicate to the user whether there is information to be referred to or not (step 1401). After that, the input information (command) using the mouse 203 is received (step 1402), and it is determined whether this input has selected the reference information output request menu 1501 or the inquiry question and answer menu.Step 1
403), if it is a reference information output request (that is, if a command to execute a program is input), the program name 12 stored in step 903
Using 01 as a key, the table 1002 is searched, the corresponding program code is extracted, loaded into the processing device 204, and executed (step 904).

In the embodiment shown above, when the system issues an inquiry to the user, only when the user is unable to answer the inquiry accurately and determines that the user would like to be presented with information to refer to, the system issues an inquiry to the user. Since related programs are started, unnecessary program starts can be avoided, which is efficient.

Furthermore, in the above-mentioned embodiment, when the information representing knowledge includes a program name, the reference information output request menu 1501 called "Reference information" is displayed on the screen. It is also possible to display the program name 1105 written inside.By doing so, multiple program names can be displayed as the reference information output request menu 1501, and the program can be executed according to the user's request. It can also be easily realized.

Of course, this program name can be entered in the query text display area 60.
It is clear that there is no particular problem with a method of displaying in accordance with No. 1.

As described above, according to the embodiments shown in FIGS. 1 to 9,
If the question content of an information inquiry during inference in a knowledge-based system includes specialized terms, explanations of the terms are displayed automatically or as needed, allowing system users to understand the target field. Even if you are an amateur, you can efficiently help people understand the content of the questions.

Further, according to the embodiments shown in FIGS. 10 to 16, it is possible to describe each event in association with information that is useful when determining the event in the knowledge expression in advance. This association is 2 normal I
Something that cannot be expressed in F-THEN format 0For example, in the event that "sales are on the rise",
The interpretation of "upward trend" varies from person to person.

It is difficult to define an upward trend in advance using a mathematical formula using specific figures for sales. In the present invention, “
A program called ``Sales Trend Graph Display'' is associated with the event ``Sales are on the rise.'' It provides users with relevant information at the level of ``reference information.'' Therefore, when an event cannot be accurately defined, related information is presented and the final decision is left to the system user.

[Effects of the Invention] As described above in detail in the embodiments, according to the present invention, when a specialized term is included in the question content of an information inquiry during inference in a knowledge-based system, an explanation of the term can be provided. Since the information is displayed automatically or as needed, even if the system user is an amateur in the subject field, the system user can be effectively encouraged to understand the content of the question.

Further, according to one aspect of the present invention, each event and information that can be used as a reference when determining the event can be described in advance in the knowledge representation in association with each other. If an event cannot be accurately defined, related information is presented and the final decision is left to the system user. In this way, in the present invention, appropriate materials and data to be referenced can be retrieved automatically or as needed and displayed on the screen in response to a query in inference, so that relevant materials and data can be displayed on the screen in order to accurately answer the query. This has the effect of saving you the trouble of searching around for information and reprocessing related data.

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing inquiry processing in the first embodiment of the present invention, FIG. 2 is a system configuration diagram in the first embodiment of the present invention, and FIG. 3 is used in the first embodiment of the present invention. FIG. 4 is a block diagram of an example of a terminology dictionary table; FIG. 4 is a diagram showing an example of rule description in the first embodiment of the present invention; FIG. 6 is a diagram showing a screen display example in the first embodiment of the present invention, FIG. 7 is a flowchart showing inquiry processing in the second embodiment of the present invention, and FIG. 8 is a diagram showing an example of screen display in the first embodiment of the present invention. FIG. 9 is a diagram showing an example of the screen display in the second embodiment of the present invention, FIG. FIG. 11 is a flowchart of guidance processing in the third embodiment of the invention.
FIG. 12 is a diagram showing an example of rule description in the third embodiment of the present invention.
14 is a diagram showing an example of a screen display in the third embodiment of the present invention. FIG. 15 is a diagram showing an example of a screen display in the third embodiment of the present invention. Flowchart of guidance processing in example, 16th
The figure is a diagram showing an example of a screen display in the fourth embodiment of the present invention. 201...Bit map display, 202
...Keyboard, 203 ...Mouse,
204...Processing device, 205...Storage device, 206...Term dictionary, 207...
Rule type knowledge memory, 208... Working memory, 209... Memory memory for cut out terms, 2
10...Query text storage area, 1002...
... Reference information storage area, 1105 ... Reference information (
program name). 30] Figure Figure Figure Figure Figure Figure Figure Figure Figure 201 Mopa 4 SF O Ray 604 Figure 20] Display Figure Figure Figure 10 Figure] Oo2 Figure Figure] 5 Figure Figure Figure 14 Figure G'1-20] Daynib, I

Claims (1)

[Claims] 1. In an inference method using a computer having an input device, an output device, a storage device, and a knowledge base, a query sentence stored in the knowledge base is output to the output device. a step of automatically detecting predetermined information in the output query text; and using the detected information to detect the predetermined information in the query text stored in the storage device. An inference method characterized by comprising the step of searching for auxiliary data related to the data and outputting it to the output device. 2. In an inference method using a computer having an input device, an output device, a storage device, and a knowledge base, the steps include: (a) outputting a query sentence stored in the knowledge base to the output device; (b) determining whether or not the predetermined information is included in the output inquiry; (c) when the result of the determination is that the predetermined information is included; (d) automatically detecting predetermined information in the output query text; (d) using the detected information to detect information related to the predetermined information in the query text stored in the storage device; searching for auxiliary data and outputting the searched auxiliary data to the output device; and (e) inputting information corresponding to the output query text and the output auxiliary data from the input device. A method of reasoning characterized by 3. The inference method according to claim 2, wherein the computer repeatedly executes the steps from (a) to (e). 4. After the step (a), and before or after the step (b), the auxiliary data stored in the storage device is calculated using predetermined information in the query. the output device includes the step of outputting inquiry information regarding whether or not to search for the search information to the output device, and the step of inputting response information to the output inquiry information from the input device, and the output response information is stored in the storage device. When the content is to search the stored auxiliary data, the following (b)
The inference method according to claim 2, characterized in that the process proceeds to step (c). 5. In an inference method using a computer having an input device, an output device, and a knowledge base, auxiliary data regarding a predetermined term in a query sentence in the knowledge base is made to correspond to the query sentence. (a) outputting the query text stored in the knowledge base to the output device; and (b) including the predetermined query text in the output query text. (c) outputting the auxiliary data stored in the knowledge base regarding the detected term in correspondence with the detected term in the output query text; (d) inputting information corresponding to the output query text and the output auxiliary data from the input device. 6. The inference method according to claim 5, wherein the computer repeatedly executes the steps from (a) to (d). 7. In an inference method using a computer having an input device, an output device, a storage device, and a knowledge base, the knowledge base stores a query sentence and a program name related to the query sentence as information representing knowledge. storing a program corresponding to the program name in the storage device; (a) outputting information representing the knowledge, including the inquiry text, stored in the knowledge base to the output device; (b) when the outputted information representing knowledge includes a program name, inputting a command for executing the program stored in the storage device corresponding to the program name from the input device; (c) executing the program in response to the input command and outputting the execution result to the output device. 8. The step (a) is characterized in that the program name included in the information representing the knowledge is output to the output device in a display format that allows it to be distinguished from the inquiry text. The inference method described in 7. 9. A knowledge base that stores knowledge including query sentences; a storage device that stores auxiliary data regarding the field to which the query sentence in the knowledge base belongs; and a storage device that stores auxiliary data regarding the field to which the query sentence in the knowledge base belongs; an output device that outputs auxiliary data related to the output, an input device that inputs information corresponding to the output query text and the output auxiliary data, and an input device that is connected to the knowledge base, the storage device, the output device, and the input device. , a computer that reads the query text from the knowledge base and transfers it to the output device, receives the input information from the input device, and reads the next query text from the knowledge base according to the received input information. An inference system characterized by: 10. A knowledge base that stores knowledge including a query text and data regarding predetermined terms in the query text in correspondence with the query text; an output device that outputs data related to terms stored in the knowledge base, an output device connected to the knowledge base, the output device, and the input device, reads the query from the knowledge base and transfers it to the output device, and outputs data related to the terms. A computer that reads out the knowledge base and transfers it to the output device, receives the input information from the input device, and reads out the next query sentence from the knowledge base according to the received input information. An inference system that 11. a knowledge base that stores therein information representing knowledge, including a query and a program name related to the query; a storage device that stores a program corresponding to the program name; an output device that outputs information representing knowledge including query sentences and execution results of the program in the storage device; an input device that inputs commands to execute the program; the knowledge base, the storage device, and the output device. and connected to the input device, reads information representing the knowledge from the knowledge base and transfers it to the output device, and receives the command from the input device when the transferred information includes a program name; An inference system comprising: a computer that reads a program from the storage device in response to the command, executes the program, and outputs the execution result of the program to the output device.