Master Thesis

Permanent URI for this collection

Now showing 1 - 1 of 1

Open Access
LLM Powered Intelligent Document Chatbot
(Mohamed Boudiaf University of M'sila, 2025-06-15) Safa, Chebbih; Sana, Amri; Supervisor: Hichem, Debbi
In the business world, documents play a foundational role for companies, especially given its influence on the company's future in the term of decision-making, business progress, financial statements, its profits, and its long-term survival in the market. With the company's growth and evaluation, employees found themselves with the need of speed and quality to efficiently handle the documents, requiring their concentration, effort, and considerable time, which occasionally leads to unsuccessful outcomes. As a solution for this problem, we proposed RAGDocAI Chat, which is a chatbot that offers a user-friendly interface that lets the employees or users effectively interact with the internal documents of their company, with comparable performance and accuracy to models such as ChatGPT and Grok, and deepseek using the power of large language models( LLMs) to generate insightful, infered answers based on the documents and the user's query in natural language.RAGDocAI also use Retrieval Augmented Generation technologies such as traditional RAG, Advanced RAG, GraphRAG, and AgenticRAG to build a knowledge domain from these documents, reducing the hallucination of the LLM due to its limited knowledge of those domains or not very familiar with them