![[2602.21598] Retrieval Challenges in Low-Resource Public Service Information: A Case Study on Food Pantry Access](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.21598] Retrieval Challenges in Low-Resource Public Service Information: A Case Study on Food Pantry Access
Summary
This article explores the challenges of retrieving public service information in low-resource environments, focusing on food pantry access. It presents an AI-powered conversational retrieval system designed to improve access to critical services.
Why It Matters
Access to public services like food pantries is crucial, especially in times of food insecurity. This research highlights the limitations of current retrieval systems in low-resource settings, emphasizing the need for improved technologies to enhance service accessibility.
Key Takeaways
- Public service information systems often suffer from fragmentation and outdated formats.
- The developed AI-powered system uses a Retrieval-Augmented Generation pipeline for better query handling.
- Pilot studies reveal significant limitations in retrieval robustness and query handling.
- The research underscores the need for advancements in conversational retrieval technologies.
- Improving access to critical public resources can significantly impact community welfare.
Computer Science > Information Retrieval arXiv:2602.21598 (cs) [Submitted on 25 Feb 2026] Title:Retrieval Challenges in Low-Resource Public Service Information: A Case Study on Food Pantry Access Authors:Touseef Hasan, Laila Cure, Souvika Sarkar View a PDF of the paper titled Retrieval Challenges in Low-Resource Public Service Information: A Case Study on Food Pantry Access, by Touseef Hasan and 2 other authors View PDF HTML (experimental) Abstract:Public service information systems are often fragmented, inconsistently formatted, and outdated. These characteristics create low-resource retrieval environments that hinder timely access to critical services. We investigate retrieval challenges in such settings through the domain of food pantry access, a socially urgent problem given persistent food insecurity. We develop an AI-powered conversational retrieval system that scrapes and indexes publicly available pantry data and employs a Retrieval-Augmented Generation (RAG) pipeline to support natural language queries via a web interface. We conduct a pilot evaluation study using community-sourced queries to examine system behavior in realistic scenarios. Our analysis reveals key limitations in retrieval robustness, handling underspecified queries, and grounding over inconsistent knowledge bases. This ongoing work exposes fundamental IR challenges in low-resource environments and motivates future research on robust conversational retrieval to improve access to critical public res...