Towards improving decision quality for rank aggregation through consensus and utility dual-driven method

Rank Aggregation Problem (RAP) aims to harness collective wisdom to offer valuable insights and support decision-making in various fields such as sociology and management. Traditional aggregation methods typically combine multiple rankings to reach a consensus. However, they often lack a critical evaluation of the quality of the aggregated permutations. This limitation hinders the application of these methods in decision-making contexts that desire high-quality outcomes. Thus, to address this gap, we propose a novel consensus and utility dual-driven two-stage aggregation method (CUDDTM) to enhance the decision quality of RAPs. In Stage 1, we aggregate rankings using the Borda method. By leveraging the hesitation strategy of Three-Way Decision (TWD), we segregate results into confident and candidate sets based on a confidence threshold. The candidate sets, considered uncertain and improvable, are subjected to reranking in the next stage, i.e., Stage 2. Stage 2 introduces two reranking rules, namely CUDDTM-I and CUDDTM-II, which combine item scores and utilities to generate the final global rank aggregation. We theoretically demonstrate that our method improves upon existing approaches without sacrificing the advantages of the Borda method. Finally, through some series of experimental analysis, we demonstrate the feasibility and effectiveness of our proposed methods in improving the decision quality of RAPs in comparison with existing methods. Our results highlight the efficacy of the hesitation strategy, the informational value in RAP, and the potential influence of manipulation on rank aggregation.