Researchers have developed a new corpus called SciTraj to better understand how scientific research evolves by analyzing citation patterns. Unlike traditional citation graphs, SciTraj categorizes citation edges into six types, linking them to specific claims within papers. This allows for a more nuanced analysis of research trajectories, revealing disciplinary siloing and identifying areas of concentrated topic emergence, particularly in computer vision and large language model-related fields. The corpus, containing over 32,000 papers and 573,000 typed citations, also supports a benchmark for predicting future research links. AI
IMPACT Provides a new tool for analyzing research trends and predicting future directions in AI subfields like NLP and LLMs.
RANK_REASON The cluster contains an academic paper detailing a new corpus and methodology for analyzing scientific research. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
