基于网站链接分析的“211工程”高校排名实证研究

An Empirical Study of Ranking "211 Project" Universities Based on Websites Hyperlinks Analysis

阐述基于网站链接分析的高校评价研究中存在的两点不足,即链接同等重要假设和数据不可靠性。然后设计一种可根据不同网站类型有侧重抓取的广度优先爬虫算法,抓取"211工程"高校网站作为研究样本。基于所得数据构成的社会网络,对社会网络分析中节点重要性测度的3种指标的排名效果进行实验,发现邻近度声望指标最优。并进一步引入PageRank思想,提出一种对邻近度声望指标进行改进的新指标UnivRank。实验结果表明,新指标的效果显著优于其他指标。

The paper points out that there are two defects in the current study on university evaluation based on websites hyperlinks analysis, the first lies in the hypothesis that all links are of equal importance and the second is the fact that the data are unreliable. Then the authors design an FBS crawling algorithm, which can crawl different types of websites to dif- ferent extent, and construct and initiate a quantitative study on the hyperlink community of universities of ＂211 Project＂. Based on the data, the authors design an experiment in order to test the three SNA indexes which can measure the significance of notes in the network. The results show that the proximity prestige index performs better than the other two indexes. Combining the proximity prestige index with PageRank arithmetic, the authors then design a new index, which is called ＂UnivRank＂ in short. The final results show that the new index performs significantly better than the other ones.