Evaluating social network reputation based on hybrid efficiency and strategy-proofness

Facing so many reputation systems in social network, the problem of evaluating a good reputation system is discussed and a solution presented. Two usual metrics applied to solve this problem are economic efficiency and strategy-proofness. However, due to the dependency of domains in real world, it is hard to measure the economic efficiency in different social networks. Thus, an alternative metric - informativeness - is introduced to approximate economic efficiency. In this paper, we propose a hybrid reputation algorithm which makes a tradeoff between informativeness and strategy-proofness to evaluate the quality of a reputation system. Experimental validation of this hybrid reputation algorithm has been conducted on real social network data from Epinions. The proposed algorithm outperforms both on the theory and experiments.