Abstract   This paper reports on customized language parsing technique for information retrieval at sentence level. As linguistic preprocess, customized language parsing is more essential for sentence retrieval than documents. Starting from natural language, it aims to extract query intention from topics and analyze relevant information within sentences. Some sentence retrieval models and query expansion strategies have been applied on the basis of the linguistic parsing results. Experiments conducted on TREC novelty track test data show that sentence retrieval with customized language parsing rank top among previously published results. It indicates that customized language parsing is more effective for sentence retrieval.

Keywords language parsing, novelty detection, sentence retrieval