Method of Consequences Inference from New Facts in Case of an Incomplete Knowledge Base

Objectives: Incomplete information contained in a knowledge base is one of the serious problems hampering the achievement of goals in the field of applied artificial intelligence. This paper provides a content-related statement and a formal description of the problem of consequences inference in view of an incomplete knowledge base. Methods/Statistical Analysis: The author proposed the original method that enables not only to detect consequences from new facts, but also determine additional facts that are needed for consequences inference. The special features of implementation of all stages of the methods are illustrated by an example. To confirm the correctness of the proposed method, a program of consequence inference in the propositional calculus has been developed. Findings: The major advantage of the suggested method is the parallel execution of disjuncts division operations in logical inference procedures. This will significantly (several times) improve the performance of inference engine implemented on the basis of this method. Application/Improvements: The method of consequences inference from new facts in case of an incomplete knowledge base can be applied to reduce the time period required for solving problems in the following areas: logical forecasting of situations development; fault detection; acquisition (accumulation and assimilation) of knowledge, etc.