SciAtlas: A Large-Scale Knowledge Graph for Automated Scientific Research

Researchers introduced SciAtlas, a large-scale heterogeneous academic knowledge graph for automated scientific discovery.

It spans over 43 million papers, 157 million entities, and 3 billion triplets across 26 disciplines.

The system also adds a neuro-symbolic retrieval method to find graph-based associations and rerank results.

By tracing structured links across fields, it aims to improve literature review, trend synthesis, and idea generation while reducing reasoning errors and cost.