Improving the Effectiveness of Local Context Analysis Based on Semantic Similarity

Local context analysis is a main way to enhance the effectiveness of query expansion in the information retrieval field. A typical query may go through a pre- refinement process to improve its retrieval power. Most of the existing local context analysis methods are attempting to solve invalid selection of additive terms, which will result in retrieval performance degradation, in the process of query expansion. In this paper, we introduce a complementary method. The new local context analysis technique is improved by incorporating semantic similarity metric into query expansion model. Finally in our experimental results, using the three groups of data sets on text retrieval conference, we show a significant enhancement of precision over current existing method in the field.