基于文本计算和链接分析的主题导航优化——以ERS网站为例

Combining Link and Content Analysis to Optimize Website Navigation——Illustrated by the Case of ERS Website

网站的信息组织是图书情报领域研究的重要内容，尤其在导航优化方面也有较多探讨。本文综合运用了文本相似度计算、链接分析、社会网络分析、聚类分析等方法，提出了兼顾页面内容和已有链接关系的网站主题导航优化方案，并针对ERS网站提出了优化的具体做法。论文通过与仅基于内容和仅基于链接的网站主题导航构建方案进行比较，证明所提方案的可行性和有效性。针对ERS网站，实例也证明通过增加语义相似度高于阈值的主题间链接，并按照相似程度对相关主题的链接进行排序，可以有效地实现主题导航优化。研究表明链接关系也是一种隐含语义关系，网站导航既要考虑语义相似度高的页面，也要考虑语义相似度低但存在链接关系的页面。

Information organization of website is an important part of Library and Information Science study, especially the website navigation guidance optimization. By using text similarity calculation, link analysis, social network analysis, and cluster analysis, this paper proposes a new method of website navigation guidance, which combines both webpage content and existing link relations, and then takes ERS Website as an example to present a detailed proposal of the optimization. Besides, this paper verifies the feasibility and effectiveness of the new method by comparing with two other methods （content-based only, and link-based only）. For ERS website, the topic-based guidance is optimized by adding links for topics that have high semantic similarity degrees above threshold, and sorting the links by topic semantic similarity degrees. The research shows that existing links imply the relevance of content. Meanwhile, both the website of high semantic similarity and the low semantic similarity with links should be taken into consideration when constructing the website guidance system.