I am proud to share the slides for “Semantic Equivalence of e-Commerce Queries”, which my eBay colleague Aritra Mandal presented this morning at the KDD 2023 Workshop on E-Commerce and Natural Language Processing (ECNLP).
Our work demonstrates how to measure e-Commerce query similarity by aggregating historical searcher behavior for frequent queries and then training a sentence transformer model that generalizes to unseen queries. We show results for proprietary eBay data as well as for the Amazon Shopping Queries Dataset used in last year’s KDD Cup.
Slides below. Enjoy!