Rough Sets Inspired Extension of Forward Inference Algorithm

The main goal of this work is to introduce theoretical background of the extended forward inference algorithm. Proposed algorithm allow to continue inference after its failure. Inference failure means that the inference engine is unable to obtain the solutions — the new facts or goals confirmation. Two-phase extension of classical inference algorithm is considered. In the first phase, classical forward inference is executed. If inference fails, second phase is activated and targeted search for additional facts is executed in the interactive mode. Inference extension proposed in this work is inspired be the rough sets theory which provides the conception of lower and upper approximations of particular sets.