Egyhealth at General Arabic Health QA (MedArabiQ): An Enhanced RAG Framework with Large-Scale Arabic Q&A Medical Data
Year: 2025
Type: conference-paper
Abstract: Arabic question-answering (Q/A) chatbots face significant challenges due to the scarcity of large, high-quality datasets and the complexities of the Arabic language, including its rich morphology, multiple dialects, and diverse writing forms.To address these challenges, we implement an enhanced retrievalaugmented generation (RAG) pipeline for Arabic medical chatbots, leveraging a dataset of approximately one million Q/A pairs collected from various Arabic healthcare resources.Experimental results demonstrate that our approach significantly outperforms previous Arabic medical QA systems, improving the quality and relevance of generated answers, with the BERTScore increasing from 0.82 to 0.86.This work represents a step forward in developing scalable and accurate Arabic medical chatbots.
Authors: Hossam Amer, Rawan Tarek Taha, Gannat Elsayed, Ensaf Hussein Mohamed
Institution: Nile University
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4416037664