In one minute

Bari, Garg, Wu, Singh and Nagel apply modern artificial intelligence to research on low energy nuclear reactions, a candidate clean energy source. They built a corpus of 3,424 documents: 2,174 PDFs collected from the LENR-CANR library's 4,743 entries plus 1,250 papers and reports gathered from other public sources. Using embedding models and topic modelling with Latent Dirichlet Allocation, BERTopic and Top2Vec, they map the structure and main themes of the field. They also release LENRsim, an experimental machine learning tool with a web interface that finds similar studies. The analysis and tools are aimed at helping researchers plan and advance their LENR studies.

Exploring artificial intelligence techniques to research low energy nuclear reactions