Study On HyperLink Analysis And Website Evaluation

This paper is the product of research project sponsored by the National Nature Science Foundation of China, which is named"The Theory, Methods and Demonstration Study of Webometrics (Certification Number: 70273032)".We discuss the hyperlink of academic website from three points of view in seven chapters. At Fist, the status quo of research are summarized in the first chapter. Then some theoretic topic on research methods, indexes, core websites and the characters of hyperlink in different layer on a website have been discussed from the second chapter to the fifth chapter. The sixth and the seventh chapter focus on the application study. The purposes of this part are not only raveling out the value of hyperlink study on science evaluation and website evaluation, but also verifying the feasibility of the research methods we established in this paper. The main ideals of each part in this paper are listed on the following parts.Chapter 1 The status quo and trends of hyperlink analysis studiesIn this chapter, we expound the significance, status quo, problems and trends of hyperlink studies. The researches of hyperlink result from the evolution of the society and the developments of network techniques, which will benefit the progress of Webometrics, exploring the cyberspace structure and knowledge mining and improving the efficiency of network information retrieval. It is a new field. The studies in theoretic field mainly involve three aspects, which are the differences between hyperlink analysis and citation analysis, the purposes and types of hyperlinks and Web-IF. The application studies mainly include measuring the core websites, exploring the cyberspace structure, improving the quality of search engine and evaluating the research ability of academic organization, et al. Although more and more researchers give highly attention to this field, and we have acquired some achievements, taking one from another, it is immature. There is not any well-rounded theory and method to use for reference, and we do not know how useful it can improve our practice either, which are the keystones in this field at present.Chapter 2 The methods of hyperlink studyAt present, all methods we used in studies on this field come from traditional methods of bibliometrics and scientometics. The unique method of this field is far from mature. In this part, we design a set of processes and develop several applications to analyze the links, which include how to sample, acquire the website data, process and analyze the data. Potential problems have been discussed, and some pieces of advice have also been given to the future studies.Chapter 3 The study on characters of hyperlinkNineteen indexes on website links character have been investigated by using twenty academic websites of business school and twenty academic websites of medical school. In this part, we draw three conclusions. Firstly, the result makes us know that measurements don't follow normal distribution. After Log transformation, the transformed measurements are roughly symmetric and unimodal. And it also indicates that there is no difference among the 101 academic website on link characters. Secondly, we find that the number of self-links is much higher than the number of outgoing-links, which can be attributed to the inconvenience of using network information resources. Thirdly, a special type of hyperlink should be given highly attention to, which is the intersection of the outgoing-links from the given host and the self-links of the given domain, because there are obviously differences in its properties among them.Chapter 4 The comparative study between the core websites and common websites on the characters of hyperlinkThe research of measuring the core websites is a hotspot in this field. In this paper, we compare two methods of measuring the core websites firstly. The result shows that there is linear relationship between them, and the coefficient is 0.75. Then the differences between these two kinds of websites on hyperlink characters are also probed into. But our conclusion is there is no difference between them because of the homogeneity of the samples. Thirdly, we established a set of reference values on the characters of academic website hyperlinks based on the descriptive statistics of the core business school websites, which are listed on the following part. We recommend that the scale of website is 964M bytes, the number of page files is 4418, the number of hyperlink is 25 each page, and the number of outgoing-links is 6 and 4 respectively on the standard of host and domain each page. These data are important for us to design a website. Fourthly, we find the distribution of number of linked websites by the samples is highly similar to Lotka distribution.Chapter 5 The study on the characters of hyperlink based on the layer of websiteThe organization and distribution of information on the website are not uniform. In this part, we study the characters of information and links in every layer of the websites, and come to several conclusions. Firstly, there is clear disciplinarian among the layers on the changes of website scale, file types and the number of hyperlinks, which have strong correlation with the function of each layer. Secondly, there are given trends of the changes on the density of hyperlink and the mean hyperlinks per page, but the diversity is great among each sample. Thirdly, the density of hyperlink and the mean hyperlinks per page are not the perfect indexes in judging the function of a page. We should better design a new index to do this. Fourthly, our research testifies that the information of all the core websites are optimized.Chapter 6 The study on the college evaluation based on college website hyperlink analysisIn this part, we discussed the relationships between college website evaluation and college evaluation. We find that there is linear relationship between Web-IF of a college website and its rank firstly. But we also find there is no linear relationship between the number of hyperlink from the other websites and the college rank. These phenomena show that the scale of website has essential influence on this kind of research. Secondly, the result indicates hyperlinks have little function on improving the rank of college, because we do not find linear relationship between the characters of website hyperlink and college rank. Thirdly, based on the above analyses, a conclusion could be drawn, that we have to do a lot of hard work in order to improve the performances of research by bring the network into play. Chapter 7 The comparative study of characters of hyperlink between Sino and American academic websitesComparing the characters of hyperlink between Sino and American academic websites, we could find that there are significant differences. The result indicates there is a huge network information gap between China and American. In order to deflate the gap, we should give highly attention to the construction of network information and do our best to promote the development of network.