Cross-language information retrieval via dictionary-based and statistics-based methods

As Internet resources become accessible to more and more countries, there is a need to develop methods for Cross Language Information Retrieval for different languages. In this paper, we focus on dictionary-based approach by using a bilingual dictionary, with a combination to statistics-based methods to avoid the problem of ambiguity. Interactive feedback loops are integrated, in the task of query expansion before and after the disambiguation of the translated candidates. In this study, we propose three sorts of query expansions to improve the effectiveness of information retrieval and to dramatically reduce the errors such an approach normally makes: an Interactive Relevance Feedback, a Domain Feedback and a Similarity Thesaurus. We applied these methods to an English-French Cross-Language Information Retrieval. In terms of average precision, a 91.95% and 99.13% of the monolingual counterpart was achieved for different combinations.