A Multi-Objective Optimization Framework for Intelligent Volunteer Matching

Abstract

Volunteer matching is not well described by conventional one-sided recommendation because effective allocation depends on reciprocal suitability, fairness, interpretability, and sustained participation. Existing studies usually address these objectives separately, and public benchmarks rarely support their joint evaluation in a volunteer-matching setting. Here, the authors propose CHM-VM, a concept- and hazard-driven framework for multi-objective volunteer matching. CHM-VM combines reciprocal representation learning, concept bottleneck matching, concept-level regularization, and hazard-aware objective routing, so that fairness, interpretability, and retention are optimized within the matching process itself. Across XING, MovieLens-1M, and MIND, CHM-VM consistently outperformed representative multi-objective, fairness-aware, interpretable, and retention-aware baselines. On XING, NDCG@10 increased from 0.403 to 0.431 and AUC from 0.744 to 0.771 relative to the strongest baseline. On MovieLens-1M, CHM-VM achieved the highest worst-group NDCG@10 (0.367) and the lowest concept disparity.