Exeter at CLEF 2002: Experiments with machine translation for monolingual and bilingual retrieval

Abstract

The University of Exeter participated in the CLEF 2002 monolingual and bilingual tasks for two languages; Italian and Spanish. Our approach for CLEF 2002 was translating the documents and topics into English, and applying our standard retrieval methods. Four ranked results were submitted for each monolingual task and five for each bilingual task. We explored the potential of term weight estimation from a contemporaneous English text collection and show that although this method has been shown to be effective in retrieval from English collections, it cannot be applied effectively to translated document collections in a simple way. This paper also reports our unofficial results for monolingual and bilingual English runs. © Springer-Verlag Berlin Heidelberg 2003.