UAE GovTrust improves bilingual public service retrieval with scope filtering
UAE GovTrust Scope Filtering Gains and Routing Failures in Bilingual Public Service Retrieval
Information Retrieval
Summary
Public service questions often mention many locations, but only one location actually controls the service needed. The authors studied a system called UAE GovTrust that helps find the right public service information in both English and Arabic. Their methods showed that filtering by the correct location scope greatly improved search accuracy. However, when the system had to decide how to route queries with tricky changes or corrections, it mostly failed. This means the system works well for ranking results but struggles with understanding complex query variations.
What this means in practice
- •For multilingual search developers: Improve ranking accuracy by incorporating location-based scope filtering in English-Arabic public service search systems.
- •For government it teams: Enhance public service portals with bilingual retrieval methods that better handle multiple location mentions for correct service matching.
Tested on one dataset.
Authors
Aleksandar Mihajlovic
Abstract
Public-service queries can mention several locations while only one controls the requested service. We study this distinction in UAE GovTrust, a controlled English-Arabic retrieval diagnostic comprising 22 authored service summaries and 50 scenarios with 110 query variants. Twelve methods separate lexical and dense ranking, rank fusion, scope filtering, a multilingual cross-encoder, and direct service lookup. On 72 matched answerable queries with bilingual evidence, role-filtered reciprocal rank fusion retrieves a relevant first result for 69 queries (95.8%), versus 47 (65.3%) for unfiltered fusion; filtered dense retrieval and direct lookup each achieve 68 (94.4%). With English-only evidence, excluding zero-score lexical candidates raises filtered fusion from 52 to 64 correct first hits. However, the unchanged router selects the correct action and scope for only 2 of 168 post-audit perturbations covering paraphrases, irrelevant facts, negation, and corrections. These perturbations are deliberately authored diagnostics, not held-out evaluation. Fresh-environment execution reproduces all 1,320 original ranking orders and actions. The results show that correct scope metadata can improve ranking while brittle scope inference can remove the evidence needed to answer. The study evaluates retrieval and routing, not generated answers or legal correctness; independent Arabic and domain validation remains necessary.