A robust rank aggregation method for malicious disturbance based on objective credit

Abstract

Rank aggregation is a task of combining individual rankings into a consensus, which has widespread applications in many areas, ranging from social choice to information retrieval. As some users may have incentives to disrupt the aggregated ranking for enormous benefits, making rank aggregation methods robust to malicious disturbance becomes a crucial challenge. In this study, we propose a robust rank aggregation method based on objective credit. The underlying idea is that a consensus ranking is obtained by combining multiple input rankings with users’ credit, while users’ credit is reflected by the differences between their input rankings and the consensus ranking. This idea motivates a novel iterative algorithm, which iteratively updates a consensus ranking weighted by users’ credit and modifies users’ credit by measuring the differences from a consensus ranking until all credit converges. In this way, the algorithm objectively assigns different credit to users, leading to a more reliable aggregated ranking. Extensive experiments on synthetic and real data demonstrate the superior performance of our method over state-of-the-art baselines.