Feasibility of Optimised Disjunctive Reasoning for Approximate Matching

Abstract. Description logics are powerful knowledge representation systems providing well-founded and computationally tractable classification reasoning. However recognition of individuals as belonging to a concept based on some approximate match to a prototypical descriptor has been a recurring application issue as description logics support only strict subsumption reasoning. Expression of concepts as a disjunction of each possible combination of sufficient prototypical features has previously been infeasible due to computational cost. Recent optimisations have greatly improved disjunctive reasoning in description logic systems and this work explores whether these are sufficient to allow the heavy use of disjunction for approximate matching. The positive results obtained support further exploration of the representation proposed within real applications.

• Type: Conference
• Year: 1999
• Where: AUSAI
• Authors: Ian Horrocks, Lin Padgham, Laura Thomson
• Keywords: Artificial Intelligence, AUSAI 1999, Description Logic, Strict Subsumption Reasoning, Tractable Classification Reasoning