FairMatch: An Explainable and Fairness-Aware Worker Recommendation Framework for Cooperative Service Marketplaces

Digital service marketplaces frequently suffer from winner-take-all dynamics where a small fraction of highly reviewed service providers receive the vast majority of recommendation exposure, locking out qualified newcomers and part-time workers. This paper formalizes, implements, and evaluates FairMatch, an explainable, multi-criteria worker recommendation framework designed for ShramSetu, a digital platform prototype for informal artisan cooperatives. FairMatch balances service suitability and distributive opportunity by combining geographic distance, customer ratings with cold-start correction, professional experience, and an inverse-workload fairness signal tracking active concurrent commitments and bookings created within the trailing 7 days that have reached completed status. Using a synthetic benchmark of 250 simulated worker profiles across five service trades and 1,000 sequential customer requests over five random seeds (5,000 total simulated transactions), FairMatch is evaluated against Nearest-First, Highest-Rated, Rating+Distance, Nearest+Least-Loaded, Round-Robin, and an ablated non-fairness baseline. Empirical results show that FairMatch improves Jain's fairness index of exposure from 0.379 ± 0.009 in the ablated model to 0.567 ± 0.017 (+49.6% relative improvement), increases booking-allocation fairness from 0.282 to 0.508 (+80.1%), and reduces exposure starvation from 86.9% (Highest-Rated) and 30.4% (ablation) to 18.0%, while maintaining a mean distance to the highest-ranked recommended worker of 2.12 km and an in-memory scoring latency of 0.008 ms. In spatial clustering evaluations where customer demand concentrates into dense urban commercial cores, Nearest-First experiences severe fairness degradation (exposure Jain dropping from 0.772 to 0.214, and workload Jain to 0.087), whereas FairMatch sustains resilient equity across exposure (0.456), bookings (0.412), and physical workload (0.322). FairMatch demonstrates how cooperative platforms can establish a defensible operational compromise between customer service proximity and equitable provider exposure.