| In the 1990's, as Internet technology develops and popularizes, academic exchange activity isn't merely restricted to paper, Internet becomes a wonderful platform which can proceed to academic exchange conveniently and immediately.There is enormous information and considerable academic information in Internet, so it's necessary that we study knowledge structure in Internet for us to know about and utilize academic website. Thus, many scholars one after another proceed to link analysis. But there is little correlative research on cositation analysis in both oversea and domestic study field.The paper regards WWW as a huge information carrier, namely the summation of technical and scientific journal, regards academic websites as technical and scientific journal, regards academic webpage as academic paper in technical and scientific journal, cites cocitation in link analysis, study reliability of using cositation to evaluate and classify academic websites. Thereby the paper enriches cositation analysis which is unsubstantial in webometrics and promotes the development of webometrics.The paper evaluates and classifies websites utilizing cositation analysis, possesses following meaning:(1) The research supplies new visual angle and method for evaluation of academic websites. At present, the method of evaluation of academic website is merely restricted to inlink analysis and web impact factor. If it is proved that there is correlation between the summation of cositation frequency of website and the star level which is evaluated by experts, it is assumed that it's feasible to evaluate the website using the summation of cositation frequency. The research supplies a new index undoubtedly for evaluation of academic website, enriches the research content of webometrics.(2) The research perfects the method of cositation analysis. Since webometrics appears, there are several scholars to proceed to cositation analysis. The paper evaluates and classifies medical academic websites using cositation analysis, in order to perfect the method of cositation analysis.(3) The research reveals the regularities of distribution of cyberspace. Owing to there is enormous academic information and websites of all kinds of discipline which is the good and bad are intermingled in network, it engenders some difficulty to use academic information. The paper can describe the knowledge structure of discipline in network to classify academic website using cositation analysis. The paper also can supply a new method to utilize academic information in network more quickly and effectively and to study knowledge structure in network.The paper analyzes the correlation between the summation of cositation frequency and the star level evaluated by experts, approaches the reliability of the summation of cositation frequency as evaluation indicator of academic websites, proves that there is significant correlation between the summation of cositation frequency and the result evaluated by experts. The summation of cositation frequency of one website is more enormous, the degree of the website in discipline is higher, academic information is more general and authority.The paper references cositation analysis method of Professor Larson in Berkeley University in USA and cocitation analysis method of Professor Cuilei in Chinese Medical University, integrates two methods, adds amendment, forms cositation analysis method in the paper. The method has five steps: first, select the website set that are studied in the research; secondly, retrieve the cositation frequency and the count of inlink of websites, construct cositation matrix, compute the summation of cositation frequency; thirdly, conform cositation matrix into similarity matrix; fourthly, use classification analysis and multidimensional scale analysis to classify the correlation matrix; fifthly, interpret cluster result. The paper analyzes 33 websites which embodied by Medical Matrix website in 9 disciplines, obtains the amount of inlink and the summation of cositation frequency, constructs cositation matrix, transforms cositation matrix into similarity matrix adopting ochiai coefficient, analyzes similarity matrix using SPSS 12.0 for Windows'classification analysis and multidimensional scale analysis, gets following important conclusion:(1) The research proved the link between academic websites isn't insignificant, it's purposeful. There is correlation between websites. The medical websites can be divided into several sub-subject under medicine using cositation analysis ideally.(2) The validity comes to 80% when multidimensional scale analysis is used to classify websites. The effect of classifying of multidimensional scale analysis is better than classification analysis.(3) In ancient research, the Boolean logic retrieve type is'link: www1 AND link: www2'. In the paper, the retrieve type is amended into'link: www1 AND NOT host: www1 AND link: www2 AND NOT host: www2'for eliminate self-link.In the process of the research, new problems appear continually, such as the matter of statistical object and sample size; the matter of precision of cositation frequency; the matter of selection of similarity; the matter of method of cositation analysis, and so on. These matters maybe impact the results a certain extent and deserve discussing and reasoning unceasingly in the following research.The research of feasibility and validity of method of evaluation and classification of websites needs systemic and numerous studies, it isn't accomplished by several dispersive and independent research. Notwithstanding the paper has limitation a certain extent, it's important. The paper enriches the evaluation method of websites and the analysis method of knowledge structure in Internet, supplies empirical data, exploits research method. No matter whether the method of cositation-based evaluation and classification of academic websites can be used in evaluation and analysis of knowledge structure of academic information in Internet or not at last, this kind of exploratory research is worthy and beneficial to do. |
PDF Full Text Request