Personalising Product Recommendations at ASOS
ASOS' recommendations are loved by its customers and are one of ASOS' core AI projects. One application of our recommender system is the You Might Also Like feature on product pages which shows similar products to customers. This talk will focus on the journey we embarked on where we aimed at personalising these product recommendations for our customers. We will talk about how we built hypotheses, how we iterated over multiple (simple and complex) models, how we performed offline and onvline tests, how we collaborated with engineers and most importantly about the successes and failures along this journey.
Key Takeaways: • Build simple models first • Iterate quickly
Sofie De Cnudde is a Machine Learning Scientist at ASOS.com. After obtaining Master’s degrees in Computer Science and Business Economics at Ghent University, she started a PhD at the University of Antwerp. Her PhD was focused on how to leverage fine-grained, human behavioural data to make predictions about people’s future actions or interests. Four publications resulted from the research where theoretical results were applied to benefit areas such as micro lending, cultural government programs and retail. She started working at ASOS in 2018 and has worked across different business areas such as Supply Chain Optimisation and Marketing, and is currently working in Recommendations.