一种基于样本加权的位置文本排序算法

为有效解决元搜索引擎结果排序的问题,提出了一种基于样本加权的位置文本排序算法。分析了现有基础算法,并在充分考虑结构信息的基础上将记录的PageRank值转化为权值。结合文本信息及其在结果列表中的位置信息实现局部相似度的调整,对排序位置的相关分值进行了规范化处理。对摘要排序算法进行了改进,提出了查询词条匹配度和词条间关联度的概念。最后将各相关分值进行合并得到搜索结果的最终相关度。实验结果表明该方法的可行性和有效性。

Position & Text Ranking Algorithm Based On Sample Weighted

To solve the result ranking of Meta Search Engine problem effectively,a text position sorting algorithm based on sample weighted was proposed.The fundamental algorithms were analyzed,full accounting the structural information was fully taken into consideration and the PageRank value was converted into weight.Combined with text information and its position in the result list,the adjustment of the local similarity was achieved,and the relevant score of position was standardized.The abstract sorting algorithm was improved,and a definition of query-match and queries relevancy was presented.The relevant scores were merged as final score at last.The experimental results show that this algorithm is feasible and efficient.