A simple way of improving traditional IR methods by structuring queries
暂无分享,去创建一个
The traditional vector-space model of information retrieval (IR) treats a query as a structureless whole, a "bag of words", thus failing to capture many finer points. We propose an algorithm that breaks a query up into smaller units, does retrieval on these, and then combines the results, thereby respecting some of the query structure. Experimental results for a standard test collection are given and discussed.
[1] Nancy Ide,et al. Introduction to the Special Issue on Word Sense Disambiguation: The State of the Art , 1998, Comput. Linguistics.
[2] Werner Winiwarter,et al. Experiments with the Use of Syntactic Analysis in Information Retrieval , 2001, NLDB.