A New Web Usage Model for Information Retrieval

Web documents are abundant, redundant and even coexist with outliers or noises. And user queries are too short and fuzzy to express their information needs. This contradictory situation brings us difficulties in matching user queries with Web documents. This paper first studies Web documents and their relative characteristics, then a new DS query term weighting scheme is proposed. Finally, a combination strategy of term weighting is given. Using this DS model, the experiments have been performed. Comparative studies indicate that the new DS term weighting improves the system performance consistently and effectively in terms of recall and precision