Information Retrieval and Recommendation Framework Based on ChatGPT

This research presents an innovative information retrieval and recommendation framework that leverages ChatGPT's capabilities to navigate the evolving landscape of language use and information growth. Focused on improving search services, the proposed ChatGPT-powered Internet data collection model integrates web crawlers for task-oriented extraction. The information retrieval algorithm uses word vectorization to address challenges in query expansion through semantic and associative matching. Collaborative filtering further improves recommendation accuracy. Carefully conducted experiments in a configured environment evaluate the framework's performance using random dataset partitioning. Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) metrics, averaged over multiple trials, measure the effectiveness of the algorithms. This research contributes a valuable framework that provides insight into recommendation algorithms within the ChatGPT paradigm, paving the way for future advances in intelligent information retrieval systems.