Trust and Distrust Prediction in Social Network with Combined Graphical and Review-Based Attributes

Trust management is of high importance in constantly growing social network systems. In particular, accurate automatic trust prediction tools can be very useful to support users of these virtual environments, especially newcomers. The aim of this work is to experimentally evaluate two groups of attributes: graph-based and those based on user ratings, in prediction of trust and distrust between a pair of users in a social network. An implementation of the C4.5 decision tree algorithm is used. The experiments are run on the real dataset: freely available, extended version of epinions.com dataset and on a wide array of attributes. The results demonstrate that a combined graph-based and review-based classifier outperforms the classifiers reported in previous related works.