AI-Driven Conversational Rental Marketplace with Fuzzy Hybrid Search and Trust-Based Moderation

Abstract

Finding rental accommodation in the Kathmandu Valley is
fragmented, broker-dependent, and plagued by fraudulent listings.
Conventional property portals reduce rich, hyper-local requirements
—water reliability, sunlight, power backup, neighborhood character
—to a handful of rigid filters, and offer little defense against fake
advertisements. This paper presents RoomieKtm, a domain-specific
rental-marketplace system that integrates conversational search,
hybrid retrieval, fuzzy re-ranking, and automated trust scoring—each
built on established techniques—into a single, purpose-built pipeline
for this market. We describe a hybrid retrieval path that fuses lexical
(BM25) and dense semantic (vector k-NN) candidates via Reciprocal
Rank Fusion, then re-ranks them with a weighted, zero-order Sugenotype
fuzzy aggregation that models graded, tolerant preferences. To
combat fraud, we introduce a five-factor trust-score model that drives
a three-way automated moderation router, complemented by a human
admin review queue, tenant reporting, and an append-only audit log.
We further describe an administrator-reviewed landlord identityverification
(KYC) workflow with explicit data-protection
safeguards. We evaluate the pipeline on a controlled pilot collection
(93 listings, 42 labeled queries) with a four-arm ablation—BM25,
semantic-only, hybrid without fuzzy re-ranking, and the full hybridplus-
fuzzy pipeline—and report paired significance tests alongside
point estimates. The fuzzy layer recovers ranking quality specifically
on paraphrased, vocabulary-mismatched queries (nDCG@10 0.662
vs. 0.622 for RRF fusion alone), and the trust router correctly flags 13
of 15 constructed fraud patterns (recall 0.867, accuracy 0.914);
however, none of the observed differences reach statistical
significance at this sample size (Wilcoxon signed-rank, all p>0.05), a
limitation we report plainly rather than obscure. We treat this as an
honest, small-scale validation of correct implementation and design
direction, not a proof of production-scale performance, and we detail
exactly what a larger, real-world evaluation would need to add.