Technique for Reducing the Number of Rules in a Temporal Knowledge Base

A knowledge base about any domain should satisfy some basic properties. One of these properties is completeness. Assuming that a knowledge base is temporal, that is, it contains explicit time references in its rules and facts, we may face a problem of an infinite number of possible premises, as the number of temporal references (e.g. in business documents) is unlimited. A technique for reducing the number of possible rules in a temporal knowledge base is proposed in this paper. This is achieved by transforming an infinite set of possible temporal rules into a finite one by mapping an infinite set of temporal references onto a finite set, and by automatic trimming of idiomatic phrases read from business documents. The technique for reducing the number of rules is presented and illustrated by an example.