Abstract:

The continuing development of the Web has led to further inaccuracy when searching across the contents. The situation is even worse when these searches are performed across different lan-guages. Cross-language information retrieval (CLIR) provides an effective way to access information re-gardless of the language in which it is authored. CLIR research has favored system-centered approaches in the past. The user is not an integral part of the translation and retrieval processes. We investigate the problem of personalized cross-language information retrieval by exploiting the results re-ranking tech-nique. The technique has been thoroughly studied in monolingual personalized information retrieval. However, the performance of results re-ranking used in personalized cross-language information retrieval is still unclear. We propose two result reranking methods for personalized cross-language information retrieval. One is based on latent semantics while another is based on external semantics. The relevant results obtained from the first retrieval round are optimized, and then we re-rank the highly relevant documents in terms of users' interests to the top of the result list. Experiments are conducted on a large real search log, and the results show that results reranking can effectively improve the search precision of personalized cross-language information retrieval.

Key words: result re-ranking, personalization, cross-language information retrieval, latent semantics, external semantics