Using a Semantic Wiki as a Knowledge Source for Rich Modeling and Question Answering

Abstract

The traditional focus of knowledge engineering has been to acquire knowledge that can be used in deductive inference. Acquiring such knowledge invariably requires users to undergo extensive training. There are, however, aspects of the knowledge necessary for deductive inference that require much less training and sophistication. For example, capturing knowledge involving ground facts and concept taxonomies requires much less training and sophistication than acquiring knowledge about deductive rules. On the basis of this observation and the recent successes of knowledge capture on the web, we are exploring the following hypothesis at the intersection between knowledge engineering and the semantic web: Can we use a semantic wiki to acquire symbolic knowledge on the web that can then be used in the context of a deductive question answering system? We are conducting this work in the context of Project Halo and the AURA system with the goal of answering Advanced Placement (AP) questions in physics, chemistry, and biology.