Oriented consensus methods for internally disjunctive production rules

Several consensus methods have been studied and modified to process internally disjunctive production rules. Oriented consensus methods have been developed which retain information concerning the intended use of production rules. The behavior of oriented consensus is studied, and an antecedent-to-antecedent consensus is defined to be a widening. Widenings can be distributed throughout a proof such that no widening occurs before any other consensus. In fact, it is shown that the widenings themselves can be distributed when the consensus is at distinct attributes. We examine the effects of preventing aU widenings from occurring during the generation of prime rules.