RecSys, the annual ACM Recommender Systems Conference held its 11th session this year in the gorgeous city of Como, Italy. As part of our platform strategy, it’s vital that we fully engage with the wider tech community, and so we brought a full team to soak up the great learnings and bring some of our own. With over 620 attendees, and a program of 46 scientific papers and 12 industry papers, the ACM Recommender Systems conference lived up to its reputation; delivering solid content, great organization and an open and pleasant atmosphere, where we presented and discussed our work with hundreds of peers.
Relive the experience in our video below.
Humberto Corona, Data Scientist at Zalando
In the last few years, some topics have started to get more and more attention within the community. This year, tourism, kids and health had their own workshops, and fashion was one of the most talked-about topics at the conference. We were delighted to be part of the conversation and share our knowledge in one of the most exciting emerging topics represented at the event.
Zalando Principal Research Engineer, Antonino Freno presented his paper showcasing our learnings and challenges when creating real world recommender systems for our large-scale platform. The paper highlights the fact that only a small part of these challenges are of an algorithmic nature. Instead, most technical problems usually arise from operational constraints, such as cost and complexity of system maintenance.
During the Large Scale Recommender Systems Workshop (LSRS), with my experience as a Research Engineer, I presented the architecture for understanding customer intent, which powers some of the personalized elements that you see in the Zalando platform. At the TempRec workshop, Sebastian Heinz a Research Scientist from Zalando Research presented our findings on using LSTM-based models for fashion recommendations, drawn from sales data of 100,000 frequent Zalando shoppers.
Fashion was heavily represented this year in general with four talks in the main conference that looked at deep learning, size recommendation, practical lessons from production systems, and outfit recommendations. The fashion recommendations community is growing and the synergy between industry and academia is getting stronger. Being part of the burgeoning industry is deeply important to us at Zalando, so we look forward to seeing what RecSys 2018 has in store.
Want to know more about our Recommender Systems team? Check out their in depth article, “Feature Extraction, Science or Engineering?”. Or to be part of our team at Zalando Tech, have a look at our jobs page. An interview with the author of this article, Humberto Corona, on his profession as a Data Scientist at Zalando is just one click away.