Abstract: The traditional sorting algoritm merely works on interlinks among web pages while putting aside the mutually reinforcing relationship between sites and pages and user evaluation of the web pages. A novel page ranking algorithm is developed to analyze update time, web-based authority and user reflected. The algorithm reckoned on a website node for each site authoritative value, in the webpage distribution authority value considered the webpage at the site location and the users of its reflection, and through the website and the webpage interact relationship to mutual feedback. The experimental results show that, compared with the traditional PageRank、HITS and other ranking algorithm, the algorithm has obvious improvement in retrieval performance.

Key words: page ranking, correlation, approval, update rate, information feedback

摘要： 传统的排序算法主要是根据网页之间的链接关系进行排序，没有考虑到网站与网页之间互相增强的关系和用户对网页的重要性的评价。为此提出了一种基于更新时间、网页权威性和用户对网页的反映的相关排序算法。该算法以网站为节点计算每个网站权威值，在为网页分配权威值时考虑了网页在网站内的位置和用户对其的反映，并通过网站与网页之间相互影响的关系来相互反馈。实验结果表明，与传统的PageRank、HITS等排序算法相比，该算法在检索性能上有明显提高。

关键词: 网页排序, 相关度, 认可度, 更新率, 信息反馈