Eliot: Interactive Tool for Exploring Fast-Changing Scientific Literature Trends
Eliot is an innovative interactive system that has been made publicly available, designed to facilitate the traceable exploration of rapidly changing scientific literature. This initiative addresses the difficulty of monitoring fast-paced research domains, inspired by investigations into Large Language Models (LLMs) and Automated Planning and Scheduling (APS). Eliot transcends traditional literature-evolution analysis by moving beyond manually created taxonomies and specific domain scripts. At query time, it retrieves arXiv papers using defined terms and filters, presents each paper with its title and abstract, categorizes the collection into themes, assigns relevant keywords, and visualizes the distribution of publications over the years. Unlike typical search engines or LLM-based tools, Eliot ensures transparent exploration of temporal trends in scientific publishing, evaluated as both a practical tool and an interactive interface for researchers to comprehend field evolution.
Key facts
- Eliot is a publicly deployed interactive system for exploring evolving scientific literature.
- It was motivated by studies on Large Language Models (LLMs) and Automated Planning and Scheduling (APS).
- Eliot retrieves arXiv papers at query time based on explicit query terms and filters.
- It represents each paper by title and abstract, clusters the corpus into themes, and assigns representative keywords.
- The system visualizes each cluster's publication-year distribution.
- Eliot generalizes literature-evolution analysis beyond hand-built taxonomies and domain-specific scripts.
- It provides traceable exploration, unlike search engines or LLM-based assistants that hide corpus selection and organization.
- The system is evaluated as both an applied system and an interactive interface.
Entities
Institutions
- arXiv