基于主题相似度模型的TS_PageRank算法_黄德才.pdf

2007 3 3 Journal of Chin s Comput r Sy st ms Vol. 28 No. 3 2007 - TS PageRank 黄德才,戚华春,钱　能 (, 310014) E-mail: hdc@ zjut. du. cn 　: Pag Rank 算法是著名有哪些信誉好的足球投注网站引擎 Googl 的核心算法,但存在主题漂移的问题,致使有哪些信誉好的足球投注网站结果中存在过多与查询主题无 的网页.在分析 Pag Rank算法及其有 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 T S-Pag Rank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 T S-Pag Rank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. : ;; Pag Rank : TP18　　　　　: A　　　　　　 : 1000-1220( 2007) 03-05 10-05 TSRank Algorithm Based on the Model of Topic Similarity HU ANG D -cai, QI Hua-chun, QIAN N ng (College of Inf ormation Engineering ,Zhejiang University of Technology , Hangzhou 310014, China) Abstract: Th Pag Rank algorith m is a k y algorithm us d in famous s arch ngin Googl , but th r x ists a bad probl m of topic- , drift which r sults in too many w b pag s without any corr lation with th us r s s arch topic in th list of w b pag s s arch d by th algorith m. Aft r analysing th Pag Rank algorithm and its modifi d algorithm , a similarity mod l bas d on virtual fil v ctor , - . - and similar d gr of cosin and put forward a T S Pag Rank algorithm f ram W can g t many diff r nt T S Pag Rank algorithms and form a s t of T S-Pag Rank algorithm, if w us diff r nt similarity mod l in th fram . Th analy sis of th ory and num rical - - simulation illustrat that th T S Pag Rank algorith m can avoid th probl m of topic drift and improv th quanlity of w b s arch ff ctiv ly without adding any oth r xtra t xt information or incr asing th d gr of tim and spac compl x ity . : ; ; Key wor