We present the QITEKAT Question-Answering system based on the conceptual theory of Knowing About Knowledge, which adopts an agent-based approach to extract information from suitable corpora. The components of the QITEKAT system entered by the School of Informatics, University of Wales, Bangor, in the 2003 Text Retrieval Conference are described in detail. We describe PPM compression techniques for Named Entity classification; distributed agent technologies for developing a Knowledgeable Framework and Knowledge Grid; and a Search Engine corroboration system for generating confidence estimates for Question Answering. We present favourable results for certain question types in the TREC Question Answering Track, and discuss future directions for the QITEKAT architecture. B.
[1]
Jennifer Chu-Carroll,et al.
A Multi-Strategy and Multi-Source Approach to Question Answering
,
2002,
TREC.
[2]
Douglas E. Appelt,et al.
FASTUS: A Cascaded Finite-State Transducer for Extracting Information from Natural-Language Text
,
1997,
ArXiv.
[3]
David J. Harper,et al.
Using compression based language models for text categorization.
,
2003
.
[4]
Eric Brill,et al.
A Simple Rule-Based Part of Speech Tagger
,
1992,
HLT.
[5]
Satoshi Sekine,et al.
Extended Named Entity Hierarchy
,
2002,
LREC.
[6]
George A. Miller,et al.
Introduction to WordNet: An On-line Lexical Database
,
1990
.