Complex verification of rule-based knowledge bases using VTF-logic

In this paper, we consider a complex procedure for verifying rule-based knowledge bases using logics with vector semantics in the form of VTF-logic with a special representation of facts and rules. The described technique allows solving verification tasks such as identifying unrelated facts, identifying incomplete rules, identifying logical circles, checking the correspondence between a set of hypotheses and a set of terminal facts, identifying contradictions, identifying silent rules, and identifying abnormal breaks in the inference chains. Rules in the knowledge base are arranged in a causal way. It means if the same fact is included in the right part of one rule and the left part of another, the first rule is always performed earlier. As a result, the verification procedure is linear in terms of the number of rules and exponential in terms of the number of starting facts. However, the amount of calculations can be significantly reduced by selecting groups of facts related to a particular hypothesis. The novelty of the study is the use of logics with vector semantics for rules verification. This retains the ability to inference in case of anomalous values of truth. This logic provides to use the inference engine for dynamic verification of knowledge. As a result, you do not need to include additional architectural elements into the system (for example, rule set dependency chart), create external verification programs, etc. In fact everything is provided by regular means of the expert system. Static verification is provided by a special representation of facts and rules.