AutoLit: Enhancing Human-AI Collaboration in Systematic Literature Reviews Through Interactive LLM Support (Stage 2, 9:59 AM)

• In Event: e-Lightning Ed-Talk Session 2 - Stage 2

Abstract

Systematic literature reviews (SLRs) are essential for identifying research gaps and synthesizing knowledge but are often labor-intensive and prone to bias. This study presents AutoLit, an interactive, LLM-powered platform designed to support researchers throughout the SLR process. AutoLit consists of four integrated modules: (1) Research Contextualization for generating backgrounds and research questions, (2) Paper Retrieval for constructing queries and retrieving literature, (3) Paper Analysis for evaluating relevance between articles and questions, and (4) Paper Coding for extracting insights. A case study was conducted to assess AutoLit's functionality, usability, and usefulness. Results demonstrate its potential to streamline SLR tasks and enhance transparency, offering practical implications for improving human-AI collaboration in academic research.

Authors

• Xiaojiao Sun, University of South Florida

• Bo Pei, University of South Florida

• Zhaowei Zhang, University of South Florida