DESIGN AND IMPLEMENTATION OF A MULTI-AGENT RAG SYSTEM FOR COLLEGE-LEVEL KNOWLEDGE PROCESSING
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This paper presents the design and implementation of a Multi-Agent Retrieval-Augmented Generation (RAG) system for college-level academic question answering. Students often face challenges in retrieving accurate and context-specific information from large academic textbooks due to limitations in traditional keyword-based search methods. To address this issue, the proposed system integrates a multiagent architecture consisting of a Query Understanding Agent, Retrieval Agent, Answer Generation Agent, and Verification Agent. The system processes academic PDF documents through a structured pipeline involving text extraction, preprocessing, chunking, embedding generation, and vector indexing using a Qdrant database. Semantic retrieval is performed using BAAI/bge-large-en-v1.5 embeddings, while answer generation is handled by the Gemma 2 9B language model, ensuring responses are grounded in retrieved context. A Streamlit-based web interface allows users to interact with the system through natural language queries, providing answers along with confidence scores and source citations. Experimental results demonstrate 90% subject detection accuracy, 84% retrieval precision, and 80% answer accuracy, with average response time below 30 seconds. The proposed system provides a scalable, reliable, and explainable solution for academic knowledge processing.
Fields of Study & Subjects
Topic Modeling
Information Retrieval and Search Behavior
Artificial Intelligence in Education
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 1 January 2026
- Language
- EN