The Improvement Of Association Strength And Correlation Coefficient In Co-citation Analysis

Since the Co-citation has been put forward by Small from 1973, its scope and approaches were expanded larger and larger. This method becomes the usual approach in Academic Research field, and was used even in Marketing Research. Co-citation Matrix is the basis of each step in the process of carrying Co-citation analysis. Meanwhile, the co-citation matrix was built according to the number that how many times the two objects were cited by another paper at the same time. The Co-citation analysis doesn't just focus on the number that the frequency of two objects were cited by another paper, but pays attention to the similarity of two objects, so we must convert the co-citation matrix to correlated matrix by using some certain similarity calculation methods. But the existed association strength calculation methods and correlation coefficient have some defects, so this paper intruded the improvement of the both index on the basis of traditional co-citation analysis system.Firstly, the paper introduced the reason why association strength calculation methods and correlation coefficient should be reformed as follows:1) the association strength between papers can't be represented by a linear trend, but be represented by exponent trend; 2) the traditional correlation coefficient usually be used to convert the data who represents at normal distribution. Secondly, the paper researched the association strength and correlation coefficient by quantitative analysis so as to prove the above two indexed need to be improved. At last, the paper evaluates the advantages and disadvantages of the new association strength and correlation coefficient by empirical studies in Information Science.The paper got some conclusions:1) the association strength between papers does represent by exponent trend; 2) During the Co-citation matrix normalization, the Spearman doesn't show its advantages. It may because the new association strength calculation method made the distinctive clearly between the objects, so the sensitivity of Spearman levels down. The improved Co-citation analysis has both advantages and defects, but it's also a valid approach in revealing the science structure, exploring the forward position of each field.