Query expansion using lexical-semantic relations

摘要: Applications such as office automation, news filtering, help facilities in complex systems, and the like require ability to retrieve documents from full-text databases where vocabulary problems can be particularly severe. Experiments performed on small collections with single-domain thesauri suggest that expanding query vectors words are lexically related original ameliorate some of mismatched vocabularies. This paper examines utility lexical expansion large, diverse TREC collection. Concepts represented by WordNet synonym sets expanded following typed links included WordNet. Experimental results show this technique makes little difference retrieval effectiveness if queries relatively complete descriptions information being sought even when concepts selected hand. Less well developed significantly improved hand-chosen concepts. However, an automatic procedure approximate set hand picked has yet devised, automatically generated degrade performance.