We look forward to welcoming you to the 15th RecSysNL meetup, hosted by bol.com in Utrecht. We'll have our usual program with one talk from academia by Sandy Manolios (TU Delft), and one from industry, by Barrie Kersbergen (bol.com).
17:30 - 18:30 Registration + Pasta Buffet
18:30 - 19:00 Intro
19:00 - 20:00 Technical Talks
20:00 - 21:00 Networking and Drinks
"The Influence of Personal Values on Music Taste: Can Personal Values be used to improve Recommendations ?"
The field of psychology can bring a lot to the field of recommender systems by adding some understanding of what drives human preference. In particular, the psychology of music highlighted the potential role of demographics, personality, social relationship and past experiences in musical preference. However, very few works investigated the potential role of personal values (what is most important for people in life) in musical preferences and its possible benefits for music recommendations. This talk will introduce a first qualitative experiment that exposes the link between those personal values and music taste.
Sandy Manolios is a Ph.D. student at TU Delft under the joint supervision of Dr. Cynthia C. S. Liem, Prof. Dr. Alan Hanjalic and Prof. Dr. Catholijn Jonker. Her main research focuses on personal values-aware music recommendations. She holds an MSc. and a BA in cognitive science and aims to bridge psychology and computer science together.
"The journey of building the Recommender systems at bol.com"
In this presentation, I will discuss the item-to-item recommender system at bol.com. How we started in 2010 and how this is maturing into a system that is now doing 2000 predictions per second on a catalog of over 70mio items. What does our architecture look like, what are the practical challenges that we faced, how we solve them and what we working on right now.
Keywords: PySpark, item-to-item, learning-to-rank, Tensorflow
Barrie Kersbergen is a senior data scientist at bol.com since 2010. He worked on the different personalization and recommender systems at bol.com.
Looking forward to seeing you all there!