“Does it come in black?” CLIP-like models are zero-shot recommenders

Patrick John Chia, Jacopo Tagliabue, Federico Bianchi, Ciro Greco, Diogo Goncalves


Abstract
Product discovery is a crucial component for online shopping. However, item-to-item recommendations today do not allow users to explore changes along selected dimensions: given a query item, can a model suggest something similar but in a different color? We consider item recommendations of the comparative nature (e.g. “something darker”) and show how CLIP-based models can support this use case in a zero-shot manner. Leveraging a large model built for fashion, we introduce GradREC and its industry potential, and offer a first rounded assessment of its strength and weaknesses.
Anthology ID:
2022.ecnlp-1.22
Volume:
Proceedings of the Fifth Workshop on e-Commerce and NLP (ECNLP 5)
Month:
May
Year:
2022
Address:
Dublin, Ireland
Venue:
ECNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
191–198
Language:
URL:
https://aclanthology.org/2022.ecnlp-1.22
DOI:
10.18653/v1/2022.ecnlp-1.22
Bibkey:
Cite (ACL):
Patrick John Chia, Jacopo Tagliabue, Federico Bianchi, Ciro Greco, and Diogo Goncalves. 2022. “Does it come in black?” CLIP-like models are zero-shot recommenders. In Proceedings of the Fifth Workshop on e-Commerce and NLP (ECNLP 5), pages 191–198, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
“Does it come in black?” CLIP-like models are zero-shot recommenders (Chia et al., ECNLP 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/ingestion-script-update/2022.ecnlp-1.22.pdf
Video:
 https://preview.aclanthology.org/ingestion-script-update/2022.ecnlp-1.22.mp4