Query translation selection for cross-language information retrieval based on HowNet

Research on cross-language information retrieval (CLIR) increasingly concentrates in candidate translation selection of the keywords in the query. The accuracy of translation has a direct impact on accurate rate and recalled rate. This thesis presents three methods based on HowNet to resolve query translation ambiguity of CLIR. The first is based on semantic relation, and it uses semantic relation network of context to determine the semantic of keywords and then select the correct translation. Bilingual decaying co-occurrence model count bilingual corpus co-occurrence information which includes the times and distance value of co-occurrence, which is different from monolingual co-occurrence. To resolve the problem of sparseness in corpus and make full use of the bilingual corpus, this paper gives another model that is semantic decaying co-occurrence model. Through test and summarizing this paper gets the best algorithm to integrate the traits of the three models, which gradually optimizes the translation and gets a higher precision.