Using neural embeddings of citation networks, researchers map scientific “disruption”: when future research stops following past directions and starts a new trajectory.

https://phys.org/news/2026-03-breakthroughs-science-large-scale-analysis.html

#Science #Innovation #Bibliometrics #MachineLearning #ResearchPolicy

A new way to detect breakthroughs in science: Large-scale analysis reveals 'disruptive' innovations in research history

The history of science and technology is marked by major breakthroughs—the theory of evolution, the splitting of the atom, the development of antibiotics—and a research team including faculty at Binghamton University, State University of New York, has developed a method to help pinpoint discoveries that reshaped the course of science.

Phys.org

@LeidenForce
Citation graphs can detect trends.
But trends are not the same as truth.

Modern science still lacks deterministic knowledge foundations.
AI measures patterns inside ambiguity.

That is the real bottleneck.
https://orcid.org/0009-0009-8968-3111

ORCID