Predicting preferred topics of authors based on co-authorship network

This paper focuses a common question in Social Network Analysis - evaluating how much a person prefers or non-prefers a specific issue. To realize this problem, we use the ILPnet2 database and model it as a co-authorship network in which the graph's nodes represent the authors and the links between two nodes means the two corresponding authors have some common papers. And what we have to do is predicting the preferred topics of authors in this network. Based on the original algorithm in [8], we propose a general algorithm with some basic assumptions and definitions and apply it to solve our problem. Finally, we use the ROC Analysis and Regression Estimation model to evaluate the Degree of Accuracy of the algorithm.