Phrase Identification in Cross-Language Information Retrieval

Term-sense ambiguity and the difficulty in translating phrases are the main sources of problem in dictionarybased cross-language information retrieval (CLIR) approaches. We propose a term similarity-based translationphrase identification technique to enhance the retrieval effectiveness of a dictionary-based query translation method. The technique identifies noun-phrases in the target language based on the degree of association between every pair of terms from two sets of translation terms. We demonstrate the effectiveness of the technique through a series of experiments using queries in two source languages, Spanish and Indonesian, to retrieve documents in English from the standard TREC (Text Retrieval Conference) collection. Combining this technique with our earlier term-similarity based sense disambiguation technique results in further retrieval performance improvements.