Smart Vendor Hub: An Evidence-Aware Adaptive Ranking and Multi-Factor Trust Framework for Hyperlocal Municipal Service Discovery

Evidence-aware ranking reduces rating manipulation and commercial bias in local service directories by conditioning vendor order on verified reviews, physical distance, and identity trust stored in a shared database index. Deploying hyperlocal discovery as a multi-role civic product requires durable relational storage, role and booking isolation, and a ranking path that does not force a separate paid-bidding or machine learning service beside the application. This paper presents a PostgreSQL-centred architecture for that setting. All vendors, service categories, and user roles share one PostgreSQL database with GIN full-text and GiST spatial indexes. Isolation and rating integrity are obtained by tagging every booking and review with customer and vendor identifiers and applying a completed-booking check before any rating updates the vendor score. Sparse ratings are stabilised by Dynamic Bayesian Shrinkage, distance is modelled by continuous Haversine decay, and identity trust is captured by a Multi-Factor Verification Trust Score (MFVTS). The design is realised in Smart Vendor Hub, an Angular 19 and FastAPI platform that connects citizens with local tradespeople, answers searches with Explainable Information Retrieval (XIR) score breakdowns, and exposes the same ranking function to a citizen discovery portal and to a live Researcher simulation lab. We describe the index schema, the ranking and review-gate algorithms, the isolation invariant, and an empirical evaluation on a deployed instance. The result is a practical blueprint for using PostgreSQL as a shared ranking store in hyperlocal service discovery without pay-to-play ad auctions or an external neural ranker.