PreSizE: Predicting Size in E-Commerce using Transformers

  title={PreSizE: Predicting Size in E-Commerce using Transformers},
  author={Yotam Eshel and Or Levi and Haggai Roitman and Alexander Nus},
  journal={Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval},
  • Yotam EshelOr Levi A. Nus
  • Published 4 May 2021
  • Computer Science, Business
  • Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval
Recent advances in the e-commerce fashion industry have led to an exploration of novel ways to enhance buyer experience via improved personalization. Predicting a proper size for an item to recommend is an important personalization challenge, and is being studied in this work. Earlier works in this field either focused on modeling explicit buyer fitment feedback or modeling of only a single aspect of the problem (e.g., specific category, brand, etc.). More recent works proposed richer models… 

Figures and Tables from this paper

Incorporating Customer Reviews in Size and Fit Recommendation systems for Fashion E-Commerce

A novel approach which can use information from customer reviews along with customer and product features for size and fit predictions is proposed which shows an improvement of 1.37% - 4.31% in F1 (macro) score over the baseline across the 4 different datasets.

Sequential Modeling with Multiple Attributes for Watchlist Recommendation in E-Commerce

The goal is to prioritize which watchlist items the user should pay attention to next by predicting the next item the user will click, and the proposed recommendation model, Trans2D, is built on top of the Transformer architecture, where it demonstrates its superiority compared to multiple state-of-the-art baselines.

Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access

This work presents a systematic review of papers on information retrieval and recommender systems that mention gender in order to document how gender is currently being used in this field and makes recommendations for ethical and well-grounded use of gender in building and researching information access systems.



A deep learning system for predicting size and fit in fashion e-commerce

This work proposes a deep learning based content-collaborative methodology for personalized size and fit recommendation that can ingest arbitrary customer and article data and can model multiple individuals or intents behind a single account.

Learning Embeddings for Product Size Recommendations

A novel size recommendation system that learns a latent space for product sizes using only past purchases and brand information is proposed, applicable to every product, without requiring specific customer or product measurements or explicit customer feedback on the purchased sizes.

Integrating Keywords into BERT4Rec for Sequential Recommendation

KeBERT4Rec is proposed, a modification of BERT4Rec, which utilizes keyword descriptions of items, and two variants for adding keywords to the model are compared on two datasets, a Movielens dataset and a dataset of an online fashion store.

Size Recommendation System for Fashion E-commerce

This paper proposes a size recommendation system to automatically pre-select consumer’s size based on past purchase and content data without explicitly asking for users measurements, and uses skip gram based word2vec model on purchase data to learn the latent representation of all products and users in a common size and fit space, thereby enabling a similarity notion among different products and user-products.

Bayesian Models for Product Size Recommendations

This work proposes a novel approach based on Bayesian logit and probit regression models with ordinal categories Small, Fit, Largeto model size fits as a function of the difference between latent sizes of customers and products, which outperforms the state-of-the-art in 5 of 6 datasets.

SSE-PT: Sequential Recommendation Via Personalized Transformer

A Personalized Transformer (SSE-PT) model, outperforming SASRec by almost 5% in terms of NDCG@10 on 5 real-world datasets, and can handle extremely long sequences and outperform SASRec in ranking results with comparable training speed, striking a balance between performance and speed requirements.

Footwear Size Recommendation System

This work presents a probabilistic approach based on user co-purchase data facilitated by generating a brand-brand relationship graph that addresses the challenge of providing recommendation for users with no previous purchases on the platform and shows significant improvement both in recommendation precision and coverage.

BERT4Rec: Sequential Recommendation with Bidirectional Encoder Representations from Transformer

A sequential recommendation model called BERT4Rec is proposed, which employs the deep bidirectional self-attention to model user behavior sequences, and outperforms various state-of-the-art sequential models consistently.

Behavior sequence transformer for e-commerce recommendation in Alibaba

This paper proposes to use the powerful Transformer model to capture the sequential signals underlying users' behavior sequences for recommendation in Alibaba and demonstrates the superiority of the proposed model.

A hierarchical bayesian model for size recommendation in fashion

A hierarchical Bayesian approach to tackle the challenging problem of size recommendation in e-commerce fashion by incorporating extended domain expertise and article characteristics as prior knowledge, which in turn makes it possible for the underlying parameters to emerge thanks to sufficient data.