JPH05158695A - Expert system - Google Patents

Abstract

PURPOSE:To replenish necessary knowledge, and restart a process at an interruption point and carry on the process if a deficiency in knowledge is detected while a knowledge process is in operation. CONSTITUTION:A knowledge process part 12 processes necessary restriction conditional expressions of knowledge codes 11 in order and specifies restriction conditional expressions to be processed; if data or restriction conditional expression is not found in the knowledge codes, the process of the restriction conditional expression is interrupted, which is reported to an interruption control part 13 to request the process. The interruption control part 13 obtains knowledge information for generating the data or restriction conditional expression to be replenished from a knowledge base 10 when the information is found in the knowledge base 10 or issues a specific input request when not to obtain the knowledge information inputted from outside, generates the necessary data or restriction conditional expression from the obtained knowledge information and adds it to the knowledge code 11; and the knowledge process part 12 is informed of that and the process of the interrupted restriction conditional expression is carried on.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an expert system for executing knowledge code, and more particularly to an expert system capable of replenishing a lack of knowledge code during execution.

[0002]

2. Description of the Related Art So-called inference processing in an expert system is performed by repeating processing of constraint condition expressions such as if eval (A, B) then exec (R). Composed.

The processing of the above-mentioned constraint condition expression is performed by the data A and B.
The value specified by "eval" is evaluated, and if the result is true (matches the condition), it means that the constraint condition expression R is processed. The number of operand data may be three or more or one depending on the content of “eval”.

Generally, the number of constraint condition expressions designated when the evaluation is true is plural, and a branch of inference occurs at the stage where a plurality of constraint condition expressions are designated, and the processing of one branch is advanced. If the evaluation result is not true in the branch where the other constraint condition expressions are stacked and the processing is advanced (that is, inference fails), the constraint condition is calculated from the latest stack point in order to proceed to another branch. Take out the expression and try the inference again.

The constraint condition expression and the data to be evaluated, which are the objects of the above processing, are collected in the knowledge base as the knowledge information representing the above contents from the required knowledge collection result, and the necessary The information is compiled into a knowledge code in an information format that is easy to process the constraint expression as described above, and is used.

FIG. 4 shows an example of the configuration of an expert system. The knowledge processing section 1 starts the processing by inputting a constraint condition expression for the user to activate necessary inference, and uses the knowledge code 2. Then, the process proceeds as described above. Knowledge code 2 is generated by separately compiling from an appropriate knowledge base.

FIG. 5 (a) shows an example of the contents of the knowledge base.
(b) is an example of the contents of the knowledge code generated by compiling (a). In the processing of the knowledge processing unit 1, if the knowledge code 2 does not have a storage area allocated to the data to be evaluated, or if there is a storage area allocated to the data but a valid value is not set ( If there is no data), or if there is no specified constraint condition expression, the process cannot proceed, so the inference process must be terminated abnormally.

Therefore, the user must judge that the knowledge of the expert system is insufficient, supplement the lacking knowledge information, compile it and add it to the knowledge code, and then restart the inference processing from the beginning.

In such a situation, especially in the test operation of the expert system under construction, the knowledge code is narrowed down to only necessary ones from the collection of enormous knowledge information, or the knowledge information It tends to occur when there is a collection omission.

An object of the present invention is to provide an expert system capable of replenishing necessary knowledge and restarting the process from the point of interruption and continuing when the lack of knowledge is detected during operation.

[0011]

FIG. 1 is a block diagram showing the configuration of the present invention. The figure shows the configuration of an expert system, which has a knowledge base 10, a knowledge code 11, a knowledge processing unit 12, and an interruption management unit 13.

The knowledge base 10 holds knowledge information in a predetermined format, and the knowledge code 11 is composed of data generated by translating the required knowledge information and constraint condition expressions, and each constraint condition expression is required. Is an expression that specifies the evaluation processing of the data and the constraint conditional expression to be processed when the result of the evaluation processing is true.

The knowledge processing unit 12 sequentially processes the required constraint condition expressions of the knowledge code, and specifies either the data or the constraint condition expressions specified in the constraint condition expression to be processed, and the knowledge code does not have any of them. In this case, the processing of the constraint conditional expression is interrupted, the interruption management unit 13 is notified, and the processing is requested.

The interruption management unit 13 acquires from the knowledge base 10 the knowledge information for generating the data or constraint condition expression that is not in the knowledge code 11 if the knowledge base 10 exists. A knowledge code is obtained by issuing a predetermined input request, externally inputting knowledge information, and generating the required data or constraint conditional expression from the acquired knowledge information.
After adding to 11, the knowledge processing unit 12 is notified and the processing of the interrupted constraint condition expression is continued.

[0015]

When the lack of knowledge is detected during system operation by the expert system of the present invention, the inference processing is temporarily interrupted, but the missing knowledge information is searched for in the knowledge base while the system operation is continued, and the knowledge information is lost. For example, it takes input from the user, compiles it, and adds it to the knowledge code.

After that, the interrupted inference processing is resumed from the interrupted status and continued, so that it becomes easy to efficiently carry out the test operation of the expert system and to prepare the knowledge code of appropriate contents.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 2 and 3 are diagrams showing an example of the processing flow of the present invention. The figure shows an example of processing one constraint condition expression, and the constraint expression is conceptually if eval (data, data, ...)
The format is then exec (constraint expression, constraint expression, ...).

When the knowledge processing unit 12 extracts one constraint condition expression from the stack in the processing step 19 of FIG. 2, the knowledge code 11 has the data for one of the data which is the operand of "eval" in the processing step 20. If yes, it is identified whether all the data has been checked in processing step 21, and if there is no check, processing returns to processing step 20 and all operands are checked in this way.

In the check, if no data is detected, the processing is interrupted and the interruption management unit 13 is requested to supplement the knowledge code. The interruption management unit 13 saves the state in order to restart the processing of the knowledge processing unit 12 in processing step 30, and then searches the knowledge base 10 for the necessary knowledge information of the data in processing step 31, and if there is, processes it. In step 32, the knowledge information is read and in step 35, it is compiled and added to the knowledge code 11.

If the knowledge base 10 does not have the necessary knowledge information, a knowledge information input request is output to the user's terminal or the like in processing step 33, and the knowledge information input entered by the user is processed in response to the request. Received at 34 and added to knowledge code 11 at process step 35 as described above. After that, the state is passed to the knowledge processing unit 12 in process step 36, and the process is restarted from the interruption point.

When the knowledge processing unit 12 finishes checking all the data in this way, the knowledge processing unit 12 performs an evaluation process of designating "eval" for these data in process step 22. If the result is true, the process shown in FIG. The process proceeds to process step 24, but if false, the process of this time is ended.

If the evaluation result is true, it is checked in the processing step 24 of FIG. 3 whether the constraint condition expression specified in the operand of "exec" exists in the knowledge code 11, and if there is, the processing step In 25, identify whether or not all specified constraint conditional expressions have been checked, and if there is no check, return to processing step 24,
Check all constraint expressions in this way.

If the knowledge code 11 does not have a necessary constraint condition expression, the processing is interrupted and the interruption management unit 13 is requested to supplement the knowledge code. The interruption management unit 13 saves the state in order to restart the processing of the knowledge processing unit 12 in processing step 40, and then, in processing steps 41 to 45 in the same manner as the processing steps 31 to 35 in FIG. After acquiring the knowledge information for compiling and adding the knowledge information to the knowledge code 11, the state is passed to the knowledge processing unit 12 in process step 46 to restart the process from the interruption point.

When the knowledge processing unit 12 finishes checking all the constraint conditional expressions in this way, it stacks the designated constraint conditional expressions in processing step 26 and finishes the processing.

[0025]

As is apparent from the above description, according to the present invention, in the knowledge processing of the expert system, when a lack of knowledge is detected during operation, necessary knowledge is supplemented and processing is performed from the point of interruption. Since it can be restarted and continued, there is a remarkable industrial effect that it becomes easy to efficiently carry out system test operations and the like, and to prepare knowledge codes of appropriate contents.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of the present invention.

FIG. 2 is a flowchart of the process of the present invention (No. 1)

FIG. 3 is a flowchart of the process of the present invention (No. 2)

FIG. 4 is a block diagram showing a conventional configuration example.

FIG. 5 is a diagram showing an example of a knowledge base and a knowledge code.

[Explanation of symbols]

 1, 12 Knowledge processing unit 2, 11 Knowledge code 10 Knowledge base 13 Interruption management unit 19-26, 30-36, 40-46 Processing steps

Claims (1)

[Claims] 1. A knowledge base (10), a knowledge code (11),
It has a knowledge processing unit (12) and an interruption management unit (13), the knowledge base (10) holds knowledge information in a predetermined format, and the knowledge code (11) stores the required knowledge information. It is composed of data generated by translation and a constraint conditional expression, and each of the constraint conditional expressions specifies the required evaluation processing of the data and the constraint conditional expression to be processed when the result of the evaluation processing is true. The knowledge processing unit (12) sequentially processes the required constraint conditional expressions of the knowledge code (11), and specifies the data and constraint conditional expressions specified in the constraint conditional expressions to be processed. If any of them is not in the knowledge code (11), the processing of the constraint condition expression is interrupted, the interruption management unit is notified and the processing is requested, and the interruption management unit (13) If the knowledge base (10) has knowledge information for generating data or constraint expression not found in (11), it is acquired from the knowledge base and Without the base, by issuing a predetermined input request,
Knowledge information is acquired by input from the outside, the required data or constraint conditional expression is generated from the acquired knowledge information, added to the knowledge code (11), and then notified to the knowledge processing unit (12). The expert system is configured to continue the processing of the interrupted constraint condition expression.