Cross lingual QA: A modular baseline in CLEF 2003

Abstract

We evaluate the feasibility of applying currently available research tools to the problem of cross lingual QA. We establish a task baseline by combining a cross lingual IR system with a monolingual QA system in a very short amount of time. A higher precision strategy involves applying the monolingual QA system to an automatically translated question, assumed to be correct. A higher coverage strategy consists of a term weighted proximity measure with varied query expansion, tuned for each individual question type. © Springer-Verlag 2004.