Skip to main content
main-content
Top

2022 | Book

Recommender Systems in Fashion and Retail

Proceedings of the Third Workshop at the Recommender Systems Conference (2021)

Editors: Prof. Nima Dokoohaki, Shatha Jaradat, Humberto Jesús Corona Pampín, Dr. Reza Shirvany

Publisher: Springer International Publishing

Book Series: Lecture Notes in Electrical Engineering

share
SHARE
insite
SEARCH

About this book

This book includes the proceedings of the third workshop on recommender systems in fashion and retail (2021), and it aims to present a state-of-the-art view of the advancements within the field of recommendation systems with focused application to e-commerce, retail, and fashion by presenting readers with chapters covering contributions from academic as well as industrial researchers active within this emerging new field. Recommender systems are often used to solve different complex problems in this scenario, such as product recommendations, size and fit recommendations, and social media-influenced recommendations (outfits worn by influencers).

Table of Contents

Frontmatter

Graph Recommendations

Using Relational Graph Convolutional Networks to Assign Fashion Communities to Users
Abstract
Community detection is a well-studied problem in machine learning and recommendation systems literature. In this paper, we study a novel variant of this problem where we assign predefined fashion communities to users in an Ecommerce ecosystem for downstream tasks. We model our problem as a link prediction task in knowledge graphs with multiple types of edges and multiple types of nodes depicting the intricate Ecommerce ecosystems. We employ Relational Graph Convolutional Networks (R-GCN) on top of this knowledge graph to determine whether a user should be assigned to a given community or not. We conduct empirical experiments on two real-world datasets from a leading fashion retailer. Our experiments demonstrate that the proposed graph-based approach performs significantly better than the non-graph-based baseline, indicating that higher order methods like GCN can improve the task of community assignment for fashion and Ecommerce users.
Amar Budhiraja, Mohak Sukhwani, Manasvi Aggarwal, Shirish Shevade, Girish Sathyanarayana, Ravindra Babu Tallamraju

Generative Recommendations

What Users Want? WARHOL: A Generative Model for Recommendation
Abstract
Current recommendation approaches help online merchants predict, for each visiting user, which subset of their existing products is the most relevant. However, besides being interested in matching users with existing products, merchants are also interested in understanding their users’ underlying preferences. This could indeed help them produce or acquire better matching products in the future. We argue that existing recommendation models cannot directly be used to predict the optimal combination of features that will make new products serve better the needs of the target audience. To tackle this, we turn to generative models, which allow us to learn explicitly distributions over product feature combinations both in text and visual space. We develop WARHOL, a product generation and recommendation architecture that takes as input past user shopping activity and generates relevant textual and visual descriptions of novel products. We show that WARHOL can approach the performance of state-of-the-art recommendation models, while being able to generate entirely new products that are relevant to the given user profiles.
Jules Samaran, Ugo Tanielian, Romain Beaumont, Flavian Vasile

Sizing and Fit Recommendations

Knowing When You Don’t Know in Online Fashion: An Uncertainty-Aware Size Recommendation Framework
Abstract
In recent years of online fashion, the availability of large-scale datasets has fueled the success of data-driven algorithmic products for supporting customers in their journey on fashion e-commerce platforms. Very often, these datasets are collected in an implicit manner, are subjective, and do not have expert annotated labels. The use of inconsistent and noisy data to train machine learning models could potentially harm their performance and generalization capabilities. In this paper, we explore uncertainty quantification metrics within the context of online size and fit recommender systems and show how they could be used to deal with noisy instances and subjective labels. We further propose an uncertainty-aware loss function based on Monte-Carlo dropout uncertainty estimation technique. Through experiments on real data at scale within the challenging domain of size and fit recommendation, we benchmark multiple uncertainty metrics and demonstrate the effectiveness of the proposed approach for training in the presence of noise.
Hareesh Bahuleyan, Julia Lasserre, Leonidas Lefakis, Reza Shirvany
SkillSF: In the Sizing Game, Your Size is Your Skill
Abstract
Rating systems are popular in the gaming industry to maximize the uncertainty of the win/loss outcome of future games by pairing players with similar skills thanks to ranking players through a score representing their “true” skill according to their performance in past games. Inspired by this approach, we propose to tackle the challenging problem of online size recommendations in fashion by reformulating a customer’s purchases as multiple sizing games where we denote (a) the return status of an article as a win/loss (size-related return) or as a draw (no size-related return) and (b) the garment/body dimensions or size as the skill. This re-framing allows us to leverage rating systems such as TrueSkill [1] and to propagate semantically meaningful information between article sizes and customer sizes. Through experimentations with real-life data, we demonstrate that our approach SkillSF is competitive with the state-of-the-art size recommendation models and offers valuable insights as to the underlying true size of customers and articles.
Hamid Zafar, Julia Lasserre, Reza Shirvany
Style-Based Interactive Eyewear Recommendations
Abstract
In this demonstration, we introduce STYLE PTTRNS, a style-based tool for eyewear recommendations. The tool first analyses the facial characteristics of a customer and matches those to eyewear characteristics in a style-harmonious manner. Consequently, the customer navigates through the eyewear collection while interacting with high-level expression attributes such as delicateness and strongness. This article presents the customer journey, our pragmatic approach to the recommendation system using styling expert input to bypass a cold-start problem and implementation details.
Michiel Braat, Jelle Stienstra

Fashion Understanding

A Critical Analysis of Offline Evaluation Decisions Against Online Results: A Real-Time Recommendations Case Study
Abstract
Offline evaluation has a widespread use in the development of recommender systems. In order to perform offline evaluation, an Information Retrieval practitioner has to make several decisions, such as, choosing metrics, train–test split strategies, true positives, how to account for biases and others. These decisions have been debated for many years and they are still open to debate today. In this work, we will trial and discuss different decisions that can be taken during offline evaluation for recommender systems. We will then compare their outcome against the results of AB tests performed in an e-commerce production system, which we consider to be the gold standard for evaluation. This is done to verify in an empirical manner which decisions present corroborate with the outcome of the results of an AB test.
Pedro Nogueira, Diogo Gonçalves, Vanessa Queiroz Marinho, Ana Rita Magalhães, João Sá
Attentive Hierarchical Label Sharing for Enhanced Garment and Attribute Classification of Fashion Imagery
Abstract
Fine-grained information extraction from fashion imagery is a challenging task due to the inherent diversity and complexity of fashion categories and attributes. Additionally, fashion imagery often depict multiple items while fashion items tend to follow hierarchical relations among various object types, categories, and attributes. In this study, we address both issues with a 2-step hierarchical deep learning pipeline consisting of (1) a low granularity object type detection module (upper body, lower body, full-body, footwear) and (2) two classification modules for garment categories and attributes based on the outcome of the first step. For the category and attribute-level classification stages, we examine a hierarchical label sharing (HLS) technique in two settings: (1) single-task learning (STL w/ HLS) and (2) multi-task learning with RNN and visual attention (MTL w/ RNN+VA). Our approach enables progressively focusing on appropriately detailed features for automatically learning the hierarchical relations of fashion and enabling predictions on images with complete outfits. Empirically, STL w/ HLS reached 93.99% top-3 accuracy while MTL w/ RNN+VA reached 97.57% top-5 accuracy for category classification on the DeepFashion benchmark, surpassing the current state of the art without requiring landmark or mask annotations nor specialized domain expertise.
Stefanos-Iordanis Papadopoulos, Christos Koutlis, Manjunath Sudheer, Martina Pugliese, Delphine Rabiller, Symeon Papadopoulos, Ioannis Kompatsiaris
Metadata
Title
Recommender Systems in Fashion and Retail
Editors
Prof. Nima Dokoohaki
Shatha Jaradat
Humberto Jesús Corona Pampín
Dr. Reza Shirvany
Copyright Year
2022
Electronic ISBN
978-3-030-94016-4
Print ISBN
978-3-030-94015-7
DOI
https://doi.org/10.1007/978-3-030-94016-4

Premium Partner