The Research Of Improved The H Type Index Of Multiple Indices

H index as a useful indicator for evaluating the scientific performance of author, has already been widely and deeply studied in the field of academic evaluation and the field of information science. H index which is a single evaluation index based on cited times has certain limitations in evaluation of authors, periodicals, research institutions and so on. Therefore the h type index of multiple indices was established, aiming at building a complete index system. This thesis focuses on an improvement of the original h type index of multiple indices. Besides cited times, download times, author’s h index and the impact factor of periodical when the paper published, and paper’s average cited times and average download times, a total of six indices, are considered in the development of the h type index of multiple indices, therefore academic influence of paper from aspects of usage, author and periodical are reflected comprehensively. At last, correlation between the h type index of multiple indices and traditional h index was investigated to verify the feasibility of multiple indices of h type index. The h type index of multiple indices is designed to make the appraisal more objective. Based on the original index, the author adds four indicators, which are paper’s cited half-life, paper’s downloaded half-life, the max annual cited number since the paper has been published and the cited number of the first three years since the paper has been published, the aim is to make the evaluation system more comprehensive and objective. Based on analysis on characteristics of the indicators, the author takes into account two indicators the max annual cited number since the paper has been published and the cited number of the first three years since the paper has been published. After using the principal component analysis module of the SPSS to calculate the weight of each index, we can get an impact score model, through which we work out scores of papers and improved h type index of multiple indices of each author. Then correlation between the improved h type index of multiple indices and the traditional h index is also investigated, and the correlation coefficient reaches0.928. That means those two indices are highly correlated and the improved h type index of multiple indices is also feasible to a certain extent. The thesis ends with comparing the improved h type index of multiple indices and the old one, the aim is to examine the difference and connection between the two indices.