Bribery in Rating Systems: A Game-Theoretic Perspective

TL;DR

This paper models bribery in online rating systems as a game, analyzing the conditions under which the system can be resistant to bribery through equilibrium analysis.

Contribution

It introduces a game-theoretic framework for understanding bribery in rating systems and identifies equilibrium conditions for bribery-proof ratings.

Findings

Existence of Nash equilibrium in bribery game

Real-world data supports the model

Conditions for bribery-proof rating systems

Abstract

Rating systems play a vital role in the exponential growth of service-oriented markets. As highly rated online services usually receive substantial revenue in the markets, malicious sellers seek to boost their service evaluation by manipulating the rating system with fake ratings. One effective way to improve the service evaluation is to hire fake rating providers by bribery. The fake ratings given by the bribed buyers influence the evaluation of the service, which further impacts the decision-making of potential buyers. In this paper, we study the bribery of a rating system with multiple sellers and buyers via a game-theoretic perspective. In detail, we examine whether there exists an equilibrium state in the market in which the rating system is expected to be bribery-proof: no bribery strategy yields a strictly positive gain. We first collect real-world data for modeling the bribery problem in rating systems. On top of that, we analyze the problem of bribery in a rating system as a static game. From our analysis, we conclude that at least a Nash equilibrium can be reached in the bribery game of rating systems.