An Information Retrieval Expansion Model Based on Wikipedia

Query expansion is one of the key technologies for improving precision and recall in information retrieval. In order to overcome limitations of single corpus, in this paper, semantic characteristics of Wikipedia corpus is combined with the standard corpus to extract more rich relationship between terms for construction of a steady Markov semantic network. Information of the entity pages and disambiguation pages in Wikipedia is comprehensively utilized to classify query terms to improve query classification accuracy. Related candidates with high quality can be used for query expansion according to semantic pruning. The proposal in our work is benefit to improve retrieval performance and to save search computational cost.