Exploring the Further Integration of Machine Translation in Multilingual Information Access

Machine Translation (MT) has been identified as a very important related technology for Multilingual Information Access (MLIA). Over the past decade, the usages of MT in MLIA are still largely concentrated on its capabilities for document translation, selection and examination. In this paper, by using a common evaluation framework, we explored the applications of MT in several unexplored or underexplored areas, which include query translation, relevance feedback, and out of vocabulary term resolution. Our experimental results demonstrate the unique contributions that MT can provide in those areas, and at the same time raise more interesting questions about how MT can be optimally integrated with MLIA