Analysis and Modeling of Effective Passage Retrieval Mechanisms in QAS

Abstract

Effective mechanisms for passage retrieval have a direct impact in the performance of question answering systems (QAS). In this paper we investigate the application of fuzzy logic models in passage retrieval (PR), and describe FuzzyPR, our passage retrieval system that applies these models. Additionally, a performance analysis of diverse mechanisms employed in passage retrieval for open domain QAS is presented. Preliminary performance results of FuzzyPR, when applied to retrieving answering passages in the TREC corpora, are provided