Abstract: Aiming at the consistency of clique behaviors, this paper proposes a collusion identification algorithm based on consistency of activity. In this algorithm, we evaluate the consistency degree of the network nodes’ behaviors by fuzzy theory for constructing similarity matrix; then, we identify colluding cliques by clustering method. The simulation results show the algorithm is effective in filtering malicious recommendation and resisting collusion attack, improves the reliability and quality of service for open networks.


Key words: colluding clique, consistency of activity, fuzzy evaluation, clustering analysis