Harnessing language models for computational literature review of emerging AI topics

Jaemin Chung, Byeongki Jeong, Young-Jae Park, Hwihun Jeong, Jiho Lee, Janghyeok Yoon, Jaewoong Choi

Abstract: Highlights•Propose a systematic computational literature review approach to effectively analyze artificial intelligence (AI) literature.•Construct a time-evolving AI landscape with AI-specific pre-trained language models and unsupervised clustering by period.•Define four quantitative metrics for assessing the degree of emergingness in AI topics— novelty, growth, coherence, and uncertainty.•Generate an interactive topic evolution map to track the historical development of AI topics.•Provide a comprehensive review of scaled dot-product attention mechanisms as a case study.

External IDs:doi:10.1016/j.ipm.2025.104245