JPS61109148A - Conflict eliminating system for production rule - Google Patents

Abstract

PURPOSE:To eliminate a conflict with high efficiency among production rules by recording history information on application frequency of rules, etc. for each rule by an application type reasoning system and using this information and meaning index information on each rule. CONSTITUTION:In case the IF parts of plural rules are coincident with a symbol character string in a short-period storage area 3 for reasoning in a rule inter- print mode, the processing is carried out to decide the THEN part of a specific rule that should be executed with priority. The using frequency and the meaning coefficient of a rule are decided from the number and the history information of the rule which is coincident with the IF part and stored to a register 6. In the same way, the information on each rule detected subsequently are stored to a register 7. The same operations are repeated successively to extract all conflicting rules out of a knowledge base 1. These extracted rules are stored to a register 8. Then the product of the using frequency and the meaning coefficient is obtained to said rules. The rule number showing the maximum value is transmitted to a reasoning mechanism.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to an inference processing method for applying knowledge technology, and in particular,
The present invention relates to a conflict resolution method for production rules during inference, etc., which is suitable for efficiently and rationally executing inference processing that involves conflict resolution when the IF portions of a plurality of rules match symbol strings at the same time.

[Background of the invention]

Conventional devices store rule usage history information such as the number of times the rule has been executed for each rule stored in the knowledge base, and update this information when using inference. but,
Conflict resolution is determined based only on usage count information.
No direct consideration was given to the semantic evaluation of rules.

[Purpose of the invention]

The purpose of the present invention is to apply rules using statistical information on past application of rules and the semantic importance of rules set for each rule, when a conflict of rules occurs in a knowledge engineering applied reasoning method, etc. The purpose of this invention is to provide a production rule conflict resolution method that can be applied to the usage environment of an inference system user by determining the priority of rules to be used.

[Summary of the invention]

For each rule using the conventional adaptive inference method.

Following the function of recording historical information such as the number of times a rule has been applied,
By using the information collected by this function and the semantic index information for each rule that is registered in advance, the priority in the case of rule conflict is calculated and the priority rule is determined. This allows you to specify the semantic importance of a rule when writing a rule, such as what is basic, what is important, what is necessary in conjunction with other rules, and can be used as effective information for conflict resolution. .

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to FIG. First, the configuration of the present invention will be explained with reference to FIGS. 1 and 2. FIG. 1 shows the overall configuration of the present invention.

Knowledge base l is a long-term storage area that stores production rules. In the knowledge base 1, as shown in FIG. 2, a rule 9, its history information register IO, and a meaning index description area 11 are stored as a set. The inference mechanism 2 performs inference processing by comparing the symbol string in the IF part of the rule in the knowledge base with the symbol string in the short-term memory area 3.

Next, the operation of the present invention will be explained using FIG.

When interpreting rules, if the IF parts of multiple rules match the symbol string in the inference short-term memory area 3, the THEN part of the rule is determined to be executed preferentially, as shown in Figure 3. Execute processing. First, I
The number of the matched rule in part F, the number of times the rule has been used from the history information, and the meaning coefficient of the rule are stored in register 6. Similarly, each piece of information about the next discovered rule is stored in 7.

The same operation is repeated one after another to extract all competing rules from knowledge base 1 and store them in 8.

For these rules, the product of the number of uses and the semantic coefficient is calculated, and the rule number indicating the maximum value is transmitted to the inference processing mechanism to execute inference processing.

〔Effect of the invention〕

According to the present invention, in contrast to conventional adaptive inference processing methods that attempt to resolve rule conflicts using only rule usage history information, it is possible to This has the effect of being able to resolve rule conflicts by taking into account the semantic importance of the rules.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the overall configuration of the present invention. FIG. 2 is an explanatory diagram showing an example of the knowledge base configuration of the present invention, and FIG. 3 is a flow diagram showing the processing procedure of the present invention. Explanation of symbols 1...Knowledge base, 2...Inference processing device, 3...
Inference short-term storage area, 4... input/output device, 5... control information storage area.

Claims (1)

[Claims] 1. A step in which knowledge is expressed in the form of production rules and inference processing is performed by comparing the stored rules in the storage area with the pattern of the entry string, and information for determining priority set in the rules is recorded. a step of setting a rule to be applied based on information; and a step of using the information for determining the priority, the frequency of use of the production rule, and the semantic importance of the production rule. Competitive solution method