JPH01173136A - Inference method - Google Patents

Abstract

PURPOSE:To realize the expression for importance of a condition term by composing a condition part of knowledge with the condition term and the reference of assurance with which the condition term is satisfied and using a means which executes a conclusion part after the condition part is satisfied when the coincidence is secured between the condition term and the fact of a working memory and the assurance of the fact satisfies the standard of the condition term. CONSTITUTION:A condition part is composed with a condition term and the reference of assurance with which the condition term is satisfied. While a conclusion part is composed with a conclusion term and the assurance showing the sureness of the conclusion term. The condition part is satisfied when the coincidence is secured between the condition term and the fact of a working memory 3 and also the assurance of the fact satisfies the reference of the condition term. Then an inference part 2 executes the conclusion part. In such a constitution, the reference of assurance satisfying the condition term is described to each condition part and then compared with the assurance of the fact for decision of genuineness of the condition part. Thus the expression is possible for the importance of the condition term.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to an inference method in the field of knowledge engineering, and particularly relates to an inference method that can express knowledge with richer expressive power.

BACKGROUND OF THE INVENTION In recent years, inference methods have attracted attention as a new problem-solving method in various fields as an application of knowledge engineering.

Conventionally, conditional judgment in inference methods involves comparing the conditional term with each fact in the working memory, and when a match is found, the conditional term is assumed to be satisfied and the conclusion part is executed.

An example of condition determination in the conventional inference method described above will be described below with reference to the drawings.

FIG. 5 shows an example of conventional knowledge representation. Fifth
In the diagram, knowledge 5 is ``Product X has finished A process,
If the B process is ready and there is no reservation for the B process, then the product X will be moved to the B process, and the confidence level is 0.8.

Here, it is assumed that the confidence level is a value between -l and 1, and the closer it is to l, the higher the certainty of the conclusion is. FIG. 3 shows an example of working memory. Each fact is represented by a fact item and a certainty bear representing the certainty of the fact.

Conditional judgment compares the conditional part and each fact in working memory. In this example, 511, 33.512, and 31.51.
Since 3 and 32 match, the condition part is satisfied and the conclusion part 521
is added to working memory. At this time, the confidence level of the fact added to the working memory is calculated based on the confidence level of the fact that matches the condition part and the confidence level of the conclusion part.

Problems to be Solved by the Invention However, the above method has a problem in that it is not possible to express the weight of each condition part in condition judgment.

In view of the above-mentioned problems, the present invention enables the condition part to describe a confidence criterion for condition judgment, and makes the condition judgment based on the criterion, thereby improving the flexibility of knowledge expression.

Means for Solving the Problems In order to solve the above problems, the inference method of the present invention uses a plurality of production rules expressed as pairs of condition parts and conclusion parts as specialized knowledge, and initial facts and additions in the inference process. In an inference method in which inference is made by comparing the working memory that temporarily stores the facts that have been obtained, and the condition part of the production rule and each fact in the working memory, the condition part is the condition part and the confidence that the condition part is satisfied. The conclusion part is made up of the conclusion part and the certainty that represents the certainty of the conclusion part, and the condition part and the fact in the working memory match, and the certainty of the fact satisfies the criteria of the condition part. It has an inference part that executes the conclusion part when the condition part is satisfied.

Effect of the present invention With the above-described configuration, the criteria for the degree of certainty that satisfies the conditional part is written in each conditional part, and the authenticity of the conditional part is determined by comparing it with the certainty of the fact. It is possible to express the degree of importance.

Embodiment Hereinafter, an inference method according to an embodiment of the present invention will be explained with reference to the drawings.

FIG. 1 shows the configuration of an inference method in an embodiment of the present invention. In FIG. 1, l is a knowledge base that expresses and stores specialized knowledge as a pair of a condition part and a conclusion part. 2 is an inference section, which includes a condition judgment section 5 and a conclusion implementation section 6.
3 is a working memory that temporarily stores initial facts and facts added to the inference process. 4 is an input/output device for inputting initial facts and displaying inference results.

The operation of the inference method configured as described above will be explained below with reference to FIGS. 1, 2, 3, and 4.

FIG. 2 shows an example of knowledge in the present invention. Second
In the diagram, knowledge m2 says, ``If product ”, the condition part 211 needs to satisfy the confidence criterion ``always'', and the condition part 212 needs to satisfy the confidence criterion ``always''.
Figure 4 is an example of a criterion-Wi inertia threshold table, where ``necessarily'' indicates that the confidence criterion "fairly" must be satisfied. A value of 1.0 corresponds to a confidence threshold of 0.2, and "fair" corresponds to a confidence threshold of 0.2.

Therefore, the condition part 211 mentioned above means that the condition part 2112 needs to be determined with a confidence level of 1.0 or more.

FIG. 3 shows an example of working memory. Each fact is represented by a pair of fact terms and a degree of certainty that represents the certainty of the fact.

The condition determining unit 5 reads one fact in the working memory 4 and one condition part of the knowledge in the knowledge base 1, and determines whether the fact and the condition part of the condition part match. If they match, the confidence threshold corresponding to the confidence criterion in the condition part is checked from the criterion-confidence threshold table 7. In this example, the confidence criterion 2111 in the condition part is a threshold value of 1.0.
corresponds to Next, it is checked whether the certainty of the fact is greater than or equal to the above-mentioned threshold, and if it is greater than or equal to the threshold, it is assumed that the conditional part is satisfied, and the next conditional part is judged. The knowledge obtained is sent to the conclusion implementation unit 6, where the confidence level is calculated based on the certainty level of the fact that matches the conditional part and the certainty level of the conclusion part, and is added to the working memory 3 together with the conclusion part. 2112 and fact 33 match, but fact 33 still has a confidence of 0.6, which is not above the threshold corresponding to criterion 2112, so knowledge 2 is not applied, even if the confidence of fact 33 is 1. It becomes applicable only when it becomes 0.

Effects of the Invention As described above, the present invention consists of a conditional part of knowledge consisting of a conditional part and a criterion of certainty that the conditional part should be satisfied, and the conditional part and the fact in working memory match, and the certainty of the fact is By providing a means for executing the conclusion part when the condition part is satisfied when the criterion of the condition part is satisfied, a knowledge expression means is provided that makes it possible to express the importance of the condition part.

[Brief explanation of the drawing]

Figure 1 is a block diagram showing the overall configuration of the present invention, Figure 2 is an explanatory diagram showing an example of knowledge in the knowledge base, Figure 3 is an explanatory diagram showing an example of working memory, and Figure 4 is a reference-confidence diagram. FIG. 5 is an explanatory diagram showing an example of a degree threshold table, and FIG. 5 is an explanatory diagram showing an example of conventional knowledge. 1...Knowledge base, 2...Inference section, 3
. . . Working memory, 5 . . . Condition judgment section, 6 . . . Conclusion execution section, 7 . . . Criteria-confidence threshold table. Name of agent: Patent attorney Toshio Nakao Figure 2: 213\"F! t' tf 37". -, Yawa5,
41, , Now Ha' (Ze V product + IT3 fluoro work))
Figure 3

Claims (1)

[Claims] A plurality of production rules expressed as pairs of condition parts and conclusion parts as specialized knowledge, a working memory that temporarily stores initial facts and facts added during the inference process, and the condition parts of the production rules and the working memory. In an inference method in which inferences are made by comparing each fact of and the conditional part is satisfied when the conditional term and the fact in the working memory match and the certainty of the fact satisfies the criterion of the conditional term, and the inference is characterized by comprising an inference part that executes the conclusion part. Method.