Fast Abductive Reasoning over Ontologies

Daniel M. Davenport, C. Ray Hill

In this paper we present a new method for reasoning abductively over instances of a triples ontology. We compute the usefulness of evidence toward making an inference rather than its truth value, probabilistic or otherwise. This allows us to process ambiguous, incomplete, and inconsistent information effectively while remaining faithful to the ontology. The method is fast, scalable, and robust. We first motivate the method with a simple cognitive model and then present details of the algorithms. Finally, we present results from experiments that indicate that the method scales well in space and time complexity and promises to be highly effective.

Subjects: 3. Automated Reasoning; 4. Cognitive Modeling


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.