文档介绍：摘要互联网的出现和迅速发展使信息检索的环境发生了重大变化,网页数量的暴涨, 使人们快速准确的获取信息的难度增大。在这种背景下,搜索引擎成了人们不可或缺的信息检索工具。但搜索引擎返回的网页数量往往很大,如何让用户在大量的返回结果中迅速找到最准确的信息,则显得尤为重要。而搜索引擎的网页相关性排序算法, 就能够决定用户是否能在排序靠前的网页中找到想要的信息,因此排序算法直接影响搜索引擎的使用体验,决定了搜索引擎的实用性。现有的搜索引擎排序算法,以基于网页链接结构的算法为主,主要的两种代表性算法是PageRank算法和HITS算法,基于这两种算法国内外许多学者和研究机构进行了不断的探索和改进。本文研究在分析国内外搜索引擎发展的基础之上,对经典排序算法进行了深入分析(如PageRank和HITS),将国内外现有的改进算法进行了归纳和总结,并针对ARC 算法(改进的HITS算法)的特点和不足,提出基于链接相似度的改进算法,并利用贝叶斯概率模型推导出简化方法。然后建立搜索引擎实验平台,对基于链接相似度的ARC 算法进行了验证,实验结果表明改进的ARC算法有效控制了主题漂移,提高了搜索引擎性能。关键字:搜索引擎排序算法川TS ARC算法主题漂移 ABSTRACT The emergence andrapiddevelopment tomake inforillationretrieval environment have undergone significant changes intheskyrocketing number ofpages,SO thatpeople accurate access toinformation more difficult,in thiscontext,search engines e anindispensable information retfieval toolforpeople,but thenumber ofpages returnedby search engines isoftenmassive,and how toget users tomassspeedy return of results urateinformation,it isparticularly search engine’S website relevancesortingalgorithm,we Calldetermine whether the user can findthe top—ranking web pages you want information,SO sortingalgorithms search engines use a directimpact on experience,determine theusefulness ofsearch search engine ranking algorithms,web—based link structure ofthealgorithm isbased,the main two kinds ofrepresentation ofthealgorithm isPageRank algorithm and the HITSalgorithm, based onthesetwoalgorithms many scholars and research institutions athome and abroad has conducted anew exploration andimprovement. Thispaper analyzes thedevelopment ofsearch engines athome andabroad,on this basis,the classicalsortingalgorithmsin-depthanalysis(such asPageRank andHITS),will beathome and abroad toimprove theexistingalgorithms are summarized forthe ARC algorithm(the improved HITS algorithm)the characteristics ings, propose link-based similarity oftheimproved algorithm,and use Bayesian probability model to derivesimplified inthe establi