Mining Answers from Texts and Knowledge Bases: Our Position
Bruce Porter, Ken Barker, James Fan, Paul Navrátil, Dan G. Tecuci, Peter Z. Yeh, Peter E. Clark · 2002
Recent advances in question answering from text have shown that information retrieval, natural language processing and machine learning techniques can go a long way in retrieving answers to certain types of questions from large bodies of text. Questions requiring more reasoning and inference, or those whose answers require synthesis or explanation are more difficult. Systems that reason over domin-specific knowledge bases are capable of more sophisticated behavior than answer retrieval systems, but are expensive in terms of their knowledge requirements. The problem of answering difficult questions from the knowledge exix’essed in text can be attacked from both ends: by improving answer retrieval from large corpora, and by making it possible for formal representations of knowledge contained in text to be authored more quickly and easily. Research Interests And Experience Our research group has interests in many aspects across the spectrum of this problem. We have experience in the knowledge representation issues involved in building large knowledge bases that capture knowledge contained in text as well as in simplifying the process of knowledge capture (Barker, Porter, and Clark 2001; Clark et al. 2001; Clark. Thompson, and Porter 2000, Clark and Porter 1997; Fan et al. 2001). We have worked in natural language generation of explanations from knowledge bases (Lester and Porter 1997) and reasoning for question answering (Rickel and Porter 1997; Clark, Thompson and Porter 1999). We have also investigated the relationship between text’s linguistic form and its meaning (Barker 1998; Barker and