AI-Augmented Literature Reviews: Efficient Clustering and Summarization for Researchers

With the collaboration between agents and large language models (LLMs), this paper presents a framework for using retrieval-based relevance search on a large collection of academic papers to facilitate exploratory literature reviews, which can subsequently be expanded into comprehensive reviews. The goal of this paper is to help researchers provide a practical, automatic pipeline, supported by a case study, to integrate agents and LLMs into the literature review process. The proposed framework consists of three key steps: large-scale document preprocessing, relevant document retrieval, and in-category document clustering and summarization. Using this framework, researchers can efficiently review large-scale academic papers in a transparent, scalable, efficient, and reproducible manner.