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.
Author(s): Nima Dokoohaki
Series: Lecture Notes in Social Networks
Publisher: Springer
Year: 2020
Language: English
Pages: 145
City: Cham
Acknowledgement
Contents
Part I Cold Start in Recommendations
Fashion Recommender Systems in Cold Start
1 Introduction
2 Techniques for Fashion Recommendation
3 Cold Start
4 Potential Solutions
4.1 Item Side Information Approaches
4.2 User Side Information Approaches
4.3 Approaches Based on Implicit Preferences
4.4 Cross-Domain Approaches
4.5 Rating Elicitation Approaches
5 Conclusion
References
Part II Complementary and Session Based Recommendation
Enabling Hyper-Personalisation: Automated Ad Creative Generation and Ranking for Fashion e-Commerce
1 Introduction
2 Related Work
3 Methodology
3.1 Automated Annotation of Photo-Shoot Images
3.1.1 Object and Person Detection
3.1.2 Fashion Category Detection
3.1.3 Gender and Face Detection
3.1.4 Scene Detection
3.1.5 Text Detection
3.2 Layout Generation
3.3 Creative Generation
3.3.1 Cropping and Scaling Input Image
3.3.2 Overlaying Text Callouts
3.4 Ranking Creatives
4 Experiments and Results
4.1 Qualitative Evaluation of Generated Layouts
4.2 Qualitative Evaluation of Cropping and Scaling Algorithm
4.2.1 Baseline Approach
4.3 Qualitative Evaluation of Generated Creatives
4.4 Evaluation of the Ranking Model
4.4.1 Evaluation Metrics
4.4.2 Quantitative Evaluation
4.4.3 Qualitative Evaluation
4.5 Evaluation of the Complete Approach
5 Applications
6 Conclusion
References
Two-Stage Session-Based Recommendations with Candidate Rank Embeddings
1 Introduction
2 Related Work
2.1 Session-Based Recommender Systems
2.2 Two-Stage Approaches
3 Problem Statement
4 Two-Stage Recommender with Candidate Rank Embeddings
4.1 The Candidate Generator
4.2 The Re-ranker
5 Experiments and Analysis
5.1 Datasets
5.2 Evaluation Metrics
5.3 Baselines
5.4 Experimental Setup
5.5 Predicting Fashion-Similar Target Clicks
5.6 Predicting the Next Click
5.7 Offline Results and Analysis
5.8 Online Experiment at Zalando
6 Conclusion and Future Work
References
Part III Outfit Recommendations
Attention-Based Fusion for Outfit Recommendation
1 Introduction
2 Related Work
3 Methodology
3.1 Common Space Fusion
3.2 Attention-Based Fusion
3.2.1 Visual Dot Product Attention
3.2.2 Stacked Visual Attention
3.2.3 Visual L-Scaled Dot Product Attention
3.2.4 Co-attention
4 Experimental Setup
4.1 Experiments and Evaluation
4.2 Baselines
4.3 Datasets
4.3.1 Polyvore68K
4.3.2 Polyvore21K
4.4 Comparison with Other Works
4.5 Training Details
5 Results
6 Conclusion
Appendix
A Dataset Item Types
References
Outfit2Vec: Incorporating Clothing Hierarchical MetaData into Outfits' Recommendation
1 Introduction
2 Related Works
3 Methodology
3.1 Methodology for Generating Outfits' Representative Vectors
3.1.1 Mapping of Item Details into Clothing Entities
3.1.2 Projecting Clothing Entities into Outfit Vectors
3.2 Outfit2Vec and PartialOutfit2Vec Models
4 Experimental Pipeline
4.1 Datasets
4.2 Whole Outfits Recommendation (Outfit2Vec)
4.3 Partial Outfits Recommendation
4.4 MultiClass Classification Evaluation
4.5 Discussion
5 Conclusions and Future Work
References
Part IV Sizing and Fit Recommendations
Learning Size and Fit from Fashion Images
1 Introduction
2 Related Work
3 Proposed Approach
3.1 Teacher-Student Learning
3.2 Statistical Modeling
3.3 SizeNet: Learning Visual Size and Fit Cues
3.3.1 Backbone Feature Extractor
3.3.2 Multi-layer Perceptron
4 Experimental Results and Discussion
4.1 Dataset
4.2 Evaluation
4.2.1 Baselines
4.2.2 Weight Importance
4.3 Brand Size Issue Scoring
4.4 Visualization of Size Issue Cues
5 Conclusion
References
Part V Generative Outfit Recommendation
Generating High-Resolution Fashion Model Images Wearing Custom Outfits
1 Introduction
2 Outfit Dataset
3 Methods
3.1 Unconditional
3.2 Conditional
4 Experiments
4.1 Unconditional
4.2 Conditional
4.3 Quantitative Results
5 Conclusion
References
