Question Answering Systems
AbstractSince the days of ``Turing test'', ``question answering'' (QA) has been a powerful paradigm in AI. Even though AI has diversified much beyond the notion of intelligent behavior proposed in the Turing test, QA remains a fundamental capability needed by a large class of systems. It is a powerful methodological tool to structure a task and specify its scope using a question grammar. The QA paradigm may not be necessarily associated with intelligent behavior. For example, QA has an interesting parallel to query processing in database systems which command a strong foothold in the commercial world. QA, however, goes much beyond what can be achieved using database systems. For example, many analytical tasks that involve gathering, correlating and analyzing information can naturally be formulated as QA problems. With the recent explosion of information available on the world-wide web, QA is a compelling framework for finding information that closely matches user needs.
In this seminar, I will talk about a QA system that is being developed under a DARPA-sponsored project on High Performance Knowledge Bases. At the heart of the system is a geo-political KB representing information about international actions, interests and agent capabilities. A first order theorem prover is used to reason with the KB with the goal of supporting an analyst in the domain of crisis management. Building this system has posed numerous issues including difficult representational choices, efficient reasoning, reformulation, merging and explanation generation. Even more difficult has been to characterize its performance and measure its capabilities. During this seminar, I will give an overview of technical problems, address one or two of them indepth, and present a prototoype demonstration.
Slideshttp://www.ai.sri.com/~chaudhri/vws (Updated Mon Jun 7 14:26:12 1999)
Text only version: http://www.ai.sri.com/~chaudhri/vws/tsld001.htm
1 Vinay Chaudhri
Top of this page