基于主题相似度模型的TS-PageRank算法

发布时间：2018-05-15 07:32

本文选题：链接分析 + 主题相似度　；参考：《小型微型计算机系统》2007年03期

【摘要】：PageRank算法是著名搜索引擎Google的核心算法,但存在主题漂移的问题,致使搜索结果中存在过多与查询主题无关的网页.在分析PageRank算法及其有关改进算法的基础上,提出了基于虚拟文档的主题相似度模型和基于主题相似度模型的TS-PageRank算法框架.只要选择不同的相似度计算模型,就可以得到不同的TS-PageRank算法,形成一个网页排序算法簇.理论分析和数值仿真实验表明,该算法在不需要额外文本信息,也不增加算法时空复杂度的情况下,就能极大地减少主题漂移现象,从而提高查询效率和质量.
[Abstract]:The PageRank algorithm is the core algorithm of the famous search engine Google, but there is the problem of topic drift, which leads to the existence of too many web pages which are independent of the query subject in the search results. Based on the analysis of the PageRank algorithm and its improved algorithm, a topic similarity model based on virtual document and a TS-PageRank algorithm framework based on topic similarity model are proposed. As long as we select different similarity calculation models, we can get different TS-PageRank algorithms and form a cluster of web page sorting algorithms. Theoretical analysis and numerical simulation experiments show that the algorithm can greatly reduce the topic drift phenomenon and improve the query efficiency and quality without the need of additional text information and the complexity of the algorithm.
【作者单位】：浙江工业大学信息工程学院浙江工业大学信息工程学院浙江工业大学信息工程学院
【基金】：浙江省自然科学基金项目(Y105118)资助.
【分类号】：TP393.01

【参考文献】