Research Of Query Expansion Technology

With the rapid development of web, the Internet is exploding with plenty of resources, but it is hard to find what we want. In the process of using search engines, one major problem is term mismatch between queries and documents. People often use different terms to describe concepts in their queries than authors use to describe the same concepts in their documents.秘笈/要诀/绝招/法宝are words used in the documents related to秘诀, which may vary from one document to another. If a user uses a query with word法宝, he/she cannot retrieve relevant documents due to the term mismatch. This is what thesis focuses on. Studies in the thesis mainly include:(1) This paper analysis the effectiveness of several categories of query expansion using term expansion or term re-weighted.(2) This paper proposes a novel method by using web-based resources for user query expansion. In this method, we download pages from the Internet, then analyze the pages and extract relevant terms Group to expand the user query. Compared with the traditional query expansion method using pre-construct static thesauri, our method can automatically construct the semantic resources according to web information. Our method has less constraint but higher efficiency.(3) This paper builds a query expansion model. In this model, the related term groups extracted from web-corpuses and the related terms extracted from document set are used in combination to improve the effectiveness of query expansion. And During the term expansion stage we adopt corresponding term select algorithm. Experiments on NTCIR-5 CLIR test set show that our method achieves an average 13.1% improvement compare to the traditional relevance feedback technique.(4) A text information retrieval experimental system has been designed and implemented and some others query expansion strategies are also implemented in the system for comparison between different strategies. Employing the system we participated in the 5th text retrieval conference (NTCIR'5), which well proved the effectiveness and the feasibility of the studies in this thesis.