Skip to main content
Disciplines
Automotive Business IT + Informatics Construction + Real Estate Electrical Engineering + Electronics Energy + Sustainability Insurance + Risk Finance + Banking Management + Leadership Marketing + Sales Mechanical Engineering + Materials
Events
DE
EN
Books
Journals
Topic Page
Marketing
Start single access now
Access for companies
EXTENDED SEARCH
Log in
Top

2020 | Book

Fashion Recommender Systems in Cold Start Read chapter Read first chapter

Fashion Recommender Systems

Editor: Prof. Nima Dokoohaki

Publisher: Springer International Publishing

Book Series : Lecture Notes in Social Networks

Part of: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Login to get access
insite
SEARCH

About this book

This book includes the proceedings of the first workshop on Recommender Systems in Fashion 2019. It presents a state of the art view of the advancements within the field of recommendation systems with focused application to e-commerce, retail and fashion. The volume covers 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 social fashion-based recommendations (outfits inspired by influencers), product recommendations, or size and fit recommendations.

The impact of social networks and the influence that fashion influencers have on the choices people make for shopping is undeniable. For instance, many people use Instagram to learn about fashion trends from top influencers, which helps them to buy similar or even exact outfits from the tagged brands in the post. When traced, customers’ social behavior can be a very useful guide for online shopping websites, providing insights on the styles the customers are really interested in, and hence aiding the online shops in offering better recommendations and facilitating customers quest for outfits.

Another well known difficulty with recommendation of similar items is the large quantities of clothing items which can be considered similar, but belong to different brands. Relying only on implicit customer behavioral data will not be sufficient in the coming future to distinguish between for recommendation that will lead to an item being purchased and kept, vs. a recommendation that might result in either the customer not following it, or eventually return the item.

Finding the right size and fit for clothes is one of the major factors not only impacting customers purchase decision, but also their satisfaction from e-commerce fashion platforms. Moreover, fashion articles have important sizing variations. Finally, customer preferences towards perceived article size and fit for their body remain highly personal and subjective which influences the definition of the right size for each customer.

The combination of the above factors leaves the customers alone to face a highly challenging problem of determining the right size and fit during their purchase journey, which in turn has resulted in having more than one third of apparel returns to be caused by not ordering the right article size. This challenge presents a huge opportunity for research in intelligent size and fit recommendation systems and machine learning solutions with direct impact on both customer satisfaction and business profitability.

Table of Contents

Frontmatter

Cold Start in Recommendations

Frontmatter
Fashion Recommender Systems in Cold Start
Abstract
With the rapid growth of online market for clothing, footwear, hairstyle, and makeup, consumers are getting increasingly overwhelmed with the volume, velocity and variety of production. Fashion Recommender Systems can tackle choice overload by suggesting the most interesting products to the users. However, recommender systems are unable to generate recommendation unless some information is collected from users. Indeed, there are situations where a recommender system is requested for recommendation while no or little information is provided by users (Cold Start problem). In this book chapter, we investigate the different scenarios where fashion recommender systems may encounter cold start problem and review approaches that have been proposed to deal with this problem. We further elaborate potential solutions that can be applied to mitigate moderate and severe cases of cold start problem.
Mehdi Elahi, Lianyong Qi

Complementary and Session Based Recommendation

Frontmatter
Enabling Hyper-Personalisation: Automated Ad Creative Generation and Ranking for Fashion e-Commerce
Abstract
Homepage is the first touch point in the customer’s journey and is one of the prominent channels of revenue for many e-commerce companies. A user’s attention is mostly captured by homepage banner images (also called Ads/Creatives). The set of banners shown and their design, influence the customer’s interest and plays a key role in optimizing the click through rates of the banners. Presently, massive and repetitive effort is put in, to manually create aesthetically pleasing banner images. Due to the large amount of time and effort involved in this process, only a small set of banners are made live at any point. This reduces the number of banners created as well as the degree of personalization that can be achieved. This paper thus presents a method to generate creatives automatically on a large scale in a short duration. The availability of diverse banners generated helps in improving personalization as they can cater to the taste of larger audience. The focus of our paper is on generating a wide variety of homepage banners that can be made as an input for a user-level personalization engine. Following are the main contributions of this paper: (1) We introduce and explain the need for large scale banner generation for e-commerce companies (2) We present on how we utilize existing deep learning based detectors which can automatically annotate the required objects/tags from the image. (3) We also propose a Genetic Algorithm based method to generate an optimal banner layout for the given image content, input components and other design constraints. (4) Further, to aid the process of picking the right set of banners, we designed a ranking method and evaluated multiple models. All our experiments have been performed on data from Myntra (http://​www.​myntra.​com), one of the top fashion e-commerce players in India.
Sreekanth Vempati, Korah T. Malayil, V. Sruthi, R. Sandeep
Two-Stage Session-Based Recommendations with Candidate Rank Embeddings
Abstract
Session-based recommender systems have gained attention recently due to their potential for providing real-time personalized recommendations with high recall, especially when compared to traditional methods like matrix factorization and item-based collaborative filtering. Two recent methods are Short-Term Attention/Memory Priority Model for Session-based Recommendation (STAMP) and Neural Attentive Session-based Recommendation (NARM). However, when we applied these two methods to the similar-item recommendation dataset of Zalando, they did not outperform a simple collaborative filtering baseline.
Aiming for improving similar-item recommendation, in this work we propose to re-rank a list of generated candidates, by employing the user session information encoded by an attention network. We confirm the efficacy of this strategy when using a novel Candidate Rank Embedding that encodes the global ranking information of each candidate in the re-ranking process. Offline and online experiments show significant improvements over the baseline in terms of recall and MRR, as well as improvements in click-through rate. Additionally, we evaluate the potential of this method on the next click prediction problem, where, when applied to STAMP and NARM, it improves recall and MRR on two publicly available real-world datasets.
José Antonio Sánchez Rodríguez, Jui-Chieh Wu, Mustafa Khandwawala

Outfit Recommendations

Frontmatter
Attention-Based Fusion for Outfit Recommendation
Abstract
This paper describes an attention-based fusion method for outfit recommendation which fuses the information in the product image and description to capture the most important, fine-grained product features into the item representation. We experiment with different kinds of attention mechanisms and demonstrate that the attention-based fusion improves item understanding. We outperform state-of-the-art outfit recommendation results on three benchmark datasets.
Katrien Laenen, Marie-Francine Moens
Outfit2Vec: Incorporating Clothing Hierarchical MetaData into Outfits’ Recommendation
Abstract
Fashion Personalisation is emerging as a major service that online retailers and brands are competing to provide. They aim to deliver more tailored recommendations to increase revenues and satisfy customers by providing them options of similar items according to their purchase history. However, many online retailers still struggle with turning customers’ data into actionable and intelligent recommendations that reflect their personalised and preferred taste of style. On the other hand due to the ever increasing use of social media, fashion brands invest in influencers’ marketing to advertise their brands to reach a larger segment of customers who strongly trust their influencers’ choices. In this context the textual and visual analysis of social media can be used to extract semantic knowledge about customers’ preferences that can be further applied in generating tailored online shopping recommendations. As style lies in the details of outfits, recommendation models should leverage the fashion metadata ranging from clothing categories and subcategories to attributes such as materials and patterns to overall style description in order to generate fine-grained recommendations. Recently, several recommendation algorithms suggested to model the latent representations of items and users with neural word embeddings approaches which showed improved results. Inspired by Paragraph Vector neural embeddings model, we present Outfit2vec and PartialOutfit2vec in which we leverage the complex relationship between user’s fashion metadata while generating outfits’ embeddings. In this paper, we also describe a methodology to generate representative vectors of hierarchically-composed fashion outfits. We evaluate our models using different strategies in comparison to the paragraph embedding models on an extensively-annotated Instagram dataset on recommendation and multi-class style classification tasks. Our models achieve better results specially in whole outfits’ ranking evaluations with an average of 22% increase.
Shatha Jaradat, Nima Dokoohaki, Mihhail Matskin

Sizing and Fit Recommendations

Frontmatter
Learning Size and Fit from Fashion Images
Abstract
Finding clothes that fit has increasingly become a hot topic in the e-commerce fashion industry. Mainly due to causing frustration on the customers side, and the large ecological and economical footprint on the companies side. Most approaches addressing this problem are based on statistical methods relying on historical data of articles purchased and returned to the store. Such approaches suffer from the cold start problem for the thousands of articles appearing on the shopping platforms every day, for which no prior purchase history is available. We propose to employ visual data to infer size and fit characteristics of fashion articles. We introduce SizeNet, a weakly-supervised teacher-student training framework that leverages the power of statistical models combined with the rich visual information from article images to learn visual cues for size and fit characteristics, capable of tackling the challenging cold start problem. We demonstrate the strong advantage of our approach through extensive experiments performed on thousands of textile garments, including dresses, trousers, knitwear, tops, etc. from hundreds of different brands.
Nour Karessli, Romain Guigourès, Reza Shirvany

Generative Outfit Recommendation

Frontmatter
Generating High-Resolution Fashion Model Images Wearing Custom Outfits
Abstract
Visualizing an outfit is an essential part of shopping for clothes. On fashion e-commerce platforms, only a limited number of outfits are visually represented, as it is impractical to photograph every possible outfit combination, even with a small assortment of garments. In this paper, we broaden the set of articles that can be combined into visualizations by training two Generative Adversarial Network (GAN) architectures on a dataset of outfits, poses, and fashion model images. Our first approach employs vanilla StyleGAN that is trained only on fashion model images. We show that this method can be used to transfer the style and the pose of one randomly generated outfit to another. In order to control the generated outfit, our second approach modifies StyleGAN by adding outfit/pose embedding networks. This enables us to generate realistic, high-resolution images of fashion models wearing a custom outfit under an input body pose.
Gökhan Yildirim, Nikolay Jetchev, Roland Vollgraf, Urs Bergmann
Metadata
Title
Fashion Recommender Systems
Editor
Prof. Nima Dokoohaki
Copyright Year
2020
Electronic ISBN
978-3-030-55218-3
Print ISBN
978-3-030-55217-6
DOI
https://doi.org/10.1007/978-3-030-55218-3

Premium Partner

    Image Credits