网页链接的主题相关性影响页面的权威性计算,传统的HITS算法仅从页面的链接结构评估页面的权威性,易导致主题漂移。对HITS算法进行了扩展,提出了一种主题驱动的HITS算法。该算法分析页面文档、链接的主题相关性,把主题相关性融入权威性计算,利用页面链接的拓扑结构传播页面的权威性。该算法能够搜索到与主题高耦合的结果,有效控制主题漂移,改善搜索质量。
The topic relevance ofhyperlink affects the authority evaluation, the conventional HITS readily leads to the topic drift occurred by the conventional method of hyperlink induction search. An algorithm of topic-driven HITS is proposed, which extends the original HITS. The presented approach accesses the topic relevance of web pages and hyperlink, employs topic relevance for estimating the authority, and propagates the authority to neighboring web pages. The given method could produce highly topic-coupling search results, dramatically decrease the topic drift, and improve the searching accuracy and quality.