No abstract

This article describes a new database—TM‐Link—that contains 12 million trademark applications and registrations across six jurisdictions. A feature of the database is the identification of trademark equivalents (or families) within and across national trademark offices. Equivalent trademarks are two, or more, insignias for the same product applied for by the same company. Unlike patents, the incentive to file for global priority is comparatively weak since legal priority for trademarks is territorial. To identify the number of true trademark equivalents we therefore create synthetic links using a neural network‐based machine learning algorithm.