Detection of dishonest behaviors in on-line networks using graph-based ranking techniques

Abstract

Dishonest behaviors in on-line networks include the problems caused by those actions performed by certain elements in a network in order to obtain some kind of benefits from the system. The analysis of this phenomenon concerns the WWW from two points of view: the Web as a collection of interrelated documents, and the social networks. In this work we study the web spam detection and the computation of trust and reputation in on-line social networks. We propose two graph-based ranking algorithms, based on different propagation models that spread the information from a set of elements in the network to compute the global relevance of all the nodes in the system.