• MODELING TRUST IN SOCIAL NETWORK

Abstract

We rely on trust in our day to day interactions and activities with each other. It is not easy to estimate it but we offer a simple and powerful method for estimating trust levels of agents in a social network using data from the agents’ reputation matrix. The reputation resultant method (RRM) is based on the mean values of the reputation rating matrix and the reputation resultant matrix. Reputation ratings are derived from the agents’ peer to peer ratings and the resultant reputation data is the relative reputation ratings by the agents. A comparison is made between the results of Singular value decomposition (SVD) and our new method, the RRM. The two methods offer results that are highly comparative with the RRM being simple, powerful and easy to understand and implement.