融合搜索引擎结果集的模糊积分算法

Fuzzy Integral Method to Merge Search Engine Results

对分布式信息检索的结果集采用模糊积分进行了融合，推导并给出了分布式信息检索的模糊积分算法．该算法可利用模糊积分的单调性，通过计算各信息源的模糊度量值采融合结果集并且评价排序效果．在实际的Web环境中针对4个搜索引擎算法进行了测试，结果发现，经模糊积分后的算法能较好地平衡合唱效应和黑马效应，并能获得更好的信息融合结果．在相同的条件下，所提算法在前100篇文档的排序中所荻得的相关文档数比Borda Count算法多3～4篇，比ComMIN算法多7～8篇．

A fuzzy integral algorithm of the distributed information retrieval is derived and given to merge the distributed information retrieval result set. The result set is merged and the evaluation is ranked by using the monotonieity of fuzzy integral and calculating the fuzzy measure values. Four different search engines are tested in the practical web environment, and the results show that the algorithm can balance chorus effect and dark horse effect better after fuzzy integral is carried out, and the information merger can be obtained well. Under same condition, on the top of 100 ranked documents, the number of documents obtained by the proposed integral algorithm is more than Borda Count algorithm 3-4 items, and ComMIN （ComMAX） algorithm 7-8 items, respectively.