Consortium Blockchain based Reputation Incentive Mechanism for Recommendation System

Abstract

Recommendation systems have been widely used in many e-commerce services, but it is difficult to gather enough participants to supply their recommendations. Moreover, participants in the system may make malicious recommendations, which will affect the accuracy of recommendation results. In order to provide better recommendation service for users, incentive mechanisms are needed to attract more participants in recommendation and curb their malicious behaviors. In this paper, we propose a consortium blockchain based reputation incentive mechanism for recommendation systems(CRIM). Firstly, the monetary rewards are used to attract participants and motivate them to take part in the recommendation. Secondly, we design the incentive mechanism with reputation which is attached to the rewards. Honest participants will gain more rewards while malicious participants will be penalized. Meanwhile, we adopt the Stackelberg game to maximize the utility of participants, and prove that the mechanism can reach a unique Nash equilibrium. Thirdly, the decentralization and immutability of blockchain can guarantee the credibility and security of the stored data, thus ensuring the openness and transparency of the recommendation. Finally, we implement the system for education resources recommendation and conduct experiments, and the results demonstrate that our incentive mechanism is effective and has significant performance when compared with other incentive mechanisms.