|LEXINGTON, KENTUCKY, USA|
|July 6-10, 2011|
Abstract: Datalog+- is a recently introduced family of Datalog-based languages, that constitutes a new framework for tractable ontology querying, and for a variety of other applications. Datalog+-extends plain Datalog by features such as existentially quantified rule heads, and, at the same time, restricts the rule bodies so to achieve decidability and tractability. Using skolemization, relevant fragments of Datalog+- can be translated into to classical logic programming with function symbols. Based on this observation, we will show how query-answering in popular tractable description logics can be performed with standard logic programming. Decidability and favorable complexity results are obtained by sound and complete methods of limiting the nesting depth of Skolem terms. This work was supported by the European Research Council under the European Community's Seventh Framework Programme (FP7/2007- 2013)/ERC grant no. 246858 DIADEM.
Abstract: IBM's Watson deep question answering system defeated two all-time champions on the Jeopardy! quiz show this February. Winning at Jeopardy! is a difficult challenge for a computer because clues are expressed in complex natural language over an extremely broad domain of topics, and because questions must be answered with very high precision and in a very short amount of time. Part of our approach to dealing with natural language involves pattern matching over the syntactic structure of the text, which we implemented using the Prolog language. In this talk I give an overview of the Watson system and discuss how our use of Prolog helped us acheive our goal of defeating the best human Jeopardy! players.