ISR Awards

Automating Author Name Disambiguation in MathSciNet

The American Mathematical Society?s MathSciNet is an academic intelligence service that provides information about reviews, abstracts, and publications in the mathematical sciences. Unlike other bibliographic data services, MathSciNet has manually curated author names in publication records, making it an authoritative information source for mathematicians worldwide. MathSciNet is facing a new challenge in name disambiguation. First, the number of author names to be disambiguated keeps increasing as research production in mathematical sciences grows at an unprecedented rate. Also, the appearance of a greater number of difficult-to-disambiguate names such as East Asian Names is accelerating in MathSciNet. Thus, this project aims to develop methods for automating author name disambiguation in MathSciNet in a way that a substantial number of author names are assigned unique author IDs without human curation, while suspicious cases go through expert inspection for high accuracy. The automated disambiguation procedure is expected to improve data quality control established by the experts and leadership team at MathSciNet.