Assigning UAT Keywords to Astronomy Literature in Science Explorer

To improve the findability and relevance of astronomy papers indexed in the Science Explorer (SciX), we are experimenting with machine-learning methods that assign keywords from the Unified Astronomy Thesaurus (UAT). The UAT supplies journal keywords for the American Astronomical Society (AAS), Publications of the Astronomical Society of the Pacific (PASP), and Icarus. However, from the earliest papers in SciX, no single system has served the astronomical community across all its journals.

As a Simple Knowledge Organization System (SKOS) controlled vocabulary, the UAT can combine alternative terms and link related concepts within astronomy, heliophysics, and planetary science. It can also promote cross-disciplinary discovery by connecting to similar concepts in thesauri of other scientific disciplines. As a community-contributed, open-source resource liberally licensed by the AAS, the UAT continuously adapts to scientific advances.

Keyword AI Labeler At Scix (KAILAS) is a version of the NASA Indus model fine-tuned on over 18,000 astronomy papers with author-assigned UAT keywords. We selected the training text consistent with all applicable publisher agreements. SciX abstract view includes current assignments as we continue to improve them based on community feedback.

At the request of NASA, the Astrophysics Data System (ADS) is expanding to serve heliophysics, planetary science, and earth science. SciX is the new multidisciplinary digital library based on ADS.