Contradiction Detection and Ontology Extension in a Never-Ending Learning System

Abstract

The notion of Contradiction is present in many aspects of the world and human information processing. As a consequence, more and more computer systems have been pushed into dealing with the contradiction detection task. Contradiction Detection (CD) is not a simple task, thus, it is subject to many discussions and approaches in different areas of human knowledge, such as Philosophy, Ethics, Linguistics, Computer Science, etc. and, as such, approached under different perspectives and goals. In this paper we focus on CD in a never-ending learning system called NELL (Never-ending Language Learner). Considering that NELL is intended to be self-supervised, as well as, self-reflective, it takes advantage of every new acquired knowledge (and stored its Knowledge Base - KB) to learn better and better each day. In this sense, NELL uses its own knowledge to achieve better performance in every new learning task. Therefore, the presence of contradictions in the KB of a never-ending learning system, like NELL, can result in the exponential propagation of incorrect knowledge that can lead to concept-drift. Following along these lines, in this work we proposed an approach to detect and eliminate contradictions from NELL’s KB. The results obtained from the performed experiments shows that the proposed approach can detect contradictions, as well as, eliminating them by deletion or by extending the KB hierarchy structure.

Authors thank the Brazilian research agency FAPESP.