基于扩散理论的HITS算法在Web挖掘中的研究与优化  

Research and optimization of HITS based on theory of diffusion in Web mining

作　　者：马瑞新[1] 邓贵仕[2] 王晓[1] 

机构地区：[1]大连理工大学软件学院,辽宁大连116621 [2]大连理工大学经济管理学部,辽宁大连116602

出　　处：《计算机应用研究》2012年第1期145-147,共3页Application Research of Computers

基　　金：国家自然基金资助项目(60803074)

摘　　要：传统的HITS算法单纯地对网页的链接结构进行分析,忽视了页面内容分析和网页的链接增幅,导致了主题偏离和搜索精度不高的问题。针对上述问题进行研究与分析,将超链接信息检索方法与页面内容相结合,根据优先情节和增长定律,提出了一种改进的基于扩散理论的HITS算法。实验结果表明,与传统的HITS和SALSA算法相比,该HITS算法能够有效地限制主题偏离,提高搜索精度,具有较高的实用价值。Traditional HITS simply analyzes the link structure of Webs while ignores the research of Web content and Web reference amplification,which results in topic deviation phenomenon.This paper in terms of the above issues to come up with a novel optimization of HITS based on the theory of diffusion.This method on the basis of priority complex and growth theorems to combine hyperlink information retrieval and Web content analysis together to improve the search accuracy of HITS.Experimental results show that compared with SALSA and the traditional HITS,the improved HITS is able to effectively restrain the topic deviation,increases the search accuracy with high practical value.

关 键 词：HITS 网页链接增幅 主题偏离 优先情节 扩散理论 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]