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

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.