Are Passages Enough? The MIRACLE Team Participation in QA@CLEF2009

Abstract

This paper summarizes the participation of the MIRACLE team in the Multilingual Question Answering Track at CLEF 2009. In this campaign, we took part in the monolingual Spanish task at ResPubliQA and submitted two runs. We have adapted our QA system to the new JRC-Acquis collection and the legal domain. We tested the use of answer filtering and ranking techniques against a baseline system using passage retrieval with no success. The run using question analysis and passage retrieval obtained a global accuracy of 0.33, while the addition of an answer filtering resulted in 0.29. We provide an analy-sis of the results for different questions types to investigate why it is difficult to leverage previous QA techniques. Another task of our work has been the appli-cation of temporal management to QA. Finally we include some discussion of the problems found with the new collection and the complexities of the domain.