ASSESSING ADVANCED RETRIEVALAUGMENTED GENERATION TECHNIQUES FOR QUESTION ANSWERING: A CASE STUDY ON GOVERNMENTAL SERVICES

Date

2025-1-15

Author

Alzariqi, Abdullah Talat Ahmed Mohammed

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

109
views

63
downloads

Cite This

This case study evaluates advanced Retrieval-Augmented Generation (RAG) methods to enhance question answering in conversational AI for government services, focusing on data from the Ministry of Health and Prevention. It assesses various RAG strategies like naive RAG, HyDE, Hybrid RAG, Corrective RAG, Self RAG, and Astute RAG to tackle issues like hallucinations and reasoning gaps. The study's methodology includes three phases: benchmarking embedding models, testing LLMs, and evaluating RAG pipelines. Results show Astute RAG excels in improving accuracy and context alignment, providing insights for optimizing AI solutions in government and healthcare, while highlighting limitations in dataset and evaluation methods.

Subject Keywords

RAG, LLM, Evaluation, Government Services

URI

https://hdl.handle.net/11511/113023

Collections

Graduate School of Informatics, Term Project