JPH03286228A - Knowledge base construction supporting device - Google Patents

Abstract

PURPOSE:To promote the efficiency of the construction of a knowledge base by providing a rule acquiring means, a means to preserve an acquired rule piece by piece, an interface means to display them and a rule constituting means capable of the combination of the fragmentary rules. CONSTITUTION:The rule is acquired by a rule acquiring part 2 to input the knowledge of an expert through an interface 1, and the fragmentary rule is acquired. Next, this fragmentary rule is inputted to a rule control part 3 while preserving the fragmentary rule in a data base 4 for preservation as it is, and is displayed in a tree display as an AND/OR tree through the interface 5. A rule constituting part 6 makes a fragmentary rule group controlled at every problem enable to be edited on a rule editing picture, and a user constitutes systematic knowledge from the fragmentary rule group through the operation of this picture. A meta-rule constituting part 8 constitutes a meta-rule from module knowledge to be inputted through the interface 5, and registers it in a base 9.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a rule-based expert system, and particularly to a support device for constructing a rule-based knowledge base used for inference therein.

[Conventional technology]

When constructing an expert system, the knowledge acquired from experts is expressed as rules, and a knowledge base is constructed so that there are no theoretical contradictions or lack of knowledge, and the goal of the expert system as a whole can be achieved. It is necessary to do so. In this case, the rules acquired from experts are fragmentary, so how can we link these fragmentary rules to the behavior we want to achieve as an expert system? In other words, how can we create a knowledge base from the fragmented knowledge of experts? The key to building an expert system is whether to build it or not. This design work is usually the job of a KE (knowledge engineer).

[Problem to be solved by the invention]

However, the task for KEs to compile a knowledge base from fragmented knowledge as described above varies widely depending on the problem handled by the expert system, and is not always easy, requiring a great deal of effort and time. The problem was that it had become necessary. SUMMARY OF THE INVENTION An object of the present invention is to provide a knowledge base construction support device that can improve the efficiency of knowledge base construction for a rule-based expert system and reduce the time and effort required.

[Means to solve the problem]

In order to achieve the above object, the knowledge base construction support device according to the present invention includes a rule acquisition means, a rule storage means for storing acquired rules in fragments, and a knowledge base construction support device according to the present invention.
an interface means for displaying saved rules;
A rule configuring means is provided that allows the displayed fragmentary rules to be combined by arbitrarily manipulating them.

[For production]

Fragmentary rules are acquired by experts inputting fragmentary knowledge, and these rules are stored as fragments. These fragmentary rules are displayed on the interface means, for example, as a tree structure showing a fragmentary link status, and the user can arbitrarily manipulate the displayed fragmentary rules using the rule configuration means. It is possible to combine them. Therefore, by combining these fragmentary rules, it is possible to easily create a logically consistent knowledge base. This supports the user in assembling the knowledge base and improves the efficiency of knowledge base construction. In other words, considering that the knowledge of experts is not necessarily systematic, but vague and fragmented, the fragmented rules obtained from that fragmented knowledge are saved for the time being, and later problems can be solved. By logically linking them when solving problems, we aim to improve the efficiency of building a knowledge base.

【Example】

Next, one embodiment of the present invention will be described in detail with reference to the drawings. In FIG. 1, expert knowledge is input through an interface 1, and rules are acquired by a rule acquisition section 2. Experts can input this knowledge as they think of it without having to think about the configuration of the expert system. In this way, fragmented rules are acquired. These fragmentary rules are stored and managed by the rule storage/management section 3, stored in a fragmentary rule storage database 4 as fragments, and displayed via the interface 5 as an AND10R tree, for example. For example, in the case of rule acquisition using a decision table, each table 21 of the decision table is created for each possible problem, and rules are created by filling in the table 21, as shown in FIG.
A rule 22 in an if-then format is acquired, and is stored and managed as is in the rule storage and management section 3, and is also displayed in a tree display 23 on the interface 5. As shown in FIG. 3, the rule configuration unit 6 allows a fragmentary rule group 31 managed for each problem to be edited on a rule editing screen 32.
By the operation in step 2, systematic knowledge can be constructed from the fragmentary rule group 31. When modularizing knowledge using meta-rules, multiple pieces of knowledge are created for each module as module knowledge. The completely defeated knowledge is registered in the knowledge base 7. The meta-rule elaboration unit 8 constructs meta-rules from conditions for executing the module knowledge input via the interface 5 and descriptions of the order of execution as priorities, etc., and registers the meta-rules in the meta-knowledge base 9. for example,
When inputting the conditions for executing module knowledge in a tabular format, as shown in FIG. become. The knowledge base constructed in this way operates as an expert system using some kind of inference mechanism.

【Effect of the invention】

According to the knowledge base construction support device of the present invention, experts can input knowledge as they come up with, so KE is not necessary at least up to the point where fragmentary rules are acquired. In addition, since fragmentary knowledge can be converted into rules and saved, it is often convenient when building a knowledge base, as seemingly unnecessary rules may prove unexpectedly useful later. be converted into Also,
It is convenient because it is possible to separately formulate the knowledge of multiple experts into rules and then combine them into a knowledge base later.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is a schematic diagram showing an example of operation in the rule acquisition section, Fig. 3 is a schematic diagram showing an example of operation in the rule composition section, and Fig. 4 is a meta-rule configuration. FIG. 1.5... Interface, 2... Rule acquisition unit, 3... Rule storage/management unit, 4... Database for storing fragment rules, 6... Rule configuration unit, 7...
Knowledge base, 8... Meta-rule component, 9... Meta-knowledge base.

Claims (1)

[Claims] (1) A rule acquisition means, a rule storage means for storing the acquired rules in fragments, an interface means for displaying the acquired and stored rules, and by arbitrarily operating the displayed fragmentary rules. A knowledge base construction support device characterized by comprising a rule configuration means that makes it possible to combine them.