A comparative analysis of embedding models for patent similarity

Grazia Sveva Ascione, Valerio Sterzi

This paper makes two contributions to the field of text-based patent similarity. First, it compares the performance of different kinds of patent-specific pretrained embedding model…