Enhancing Enterprise Retrieval-Augmented Generation Systems through Reinforcement Learning-Based Adaptive Retrieval
Year: 2024
Type: article
Abstract: This study suggests an adaptive retrieval framework based on reinforcement learning for analyzing enterprise Retrieval-Augmented Generation (RAG) systems. In information retrieval systems, Deep Q-Networks are used to optimize document ranking in order to determine a relevance and accuracy of retrieved documents. Experimental results show better performance than traditional BM25 and Dense Retrieval. The framework can effectively improve the accessibility of enterprise knowledge and the efficiency of enterprise systems, through continuous learning and optimized reward, providing the activity of the enterprise environment with robust and intelligent information retrieval capabilities.
Author: Hari Krishna Pokala
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217308862
