Cross-language information retrieval (CLIR), where queries and retrieved documents are in different languages, has of late become one of the major topics within the information retrieval and natural language processing communities. This paper proposes a Japanese/English CLIR system focusing mainly on the translation of technical terms. Since technical terms, most of which are compound words, can be progressively generated simply by combining existing base words, it is not entirely satisfactory or feasible to exhaustively enumerate newly emerging terms in translation dictionaries. To counter this problem, we use a bilingual dictionary of base words and collocational statistics to resolve translation ambiguity. Experiments using the NACSIS test collection showed that our system is quite comparable with a monolingual IR system in performance.
展开▼