SoundCloud is excited to host the February installment of the Recommendation Systems Stammtisch.
Join us on the 27th for an evening of mirth, pizza, RecSys talks, and more mirth.
Doors are open at 19:00.
Özgür Demir - SoundCloud - Recommendations at SoundCloud
Since its foundation SoundCloud has become one of the major platforms for user generated audio content. The uniqueness of the uploaded content together with its sheer mass makes it very difficult for the enduser to find relevant content. Hence, fully automated recommendations become a crucial part of an outstanding user experience. At SoundCloud currently various projects deal with personalized as well as unpersonalized recommendations and its related topics e.g. content classification. This talk will give a brief overview about those projects and the used technologies.
Alexandros Karatzoglou - Telefonica Research - Ranking and Diversity in Recommendations (also meet Okapi)
Most users only pay attention to the top 5 to 10 recommendations (on Mobile domains even less) it is thus very important to get these recommendations right. Ranking algorithms can help achieve this by using most of the modelling power to get the most relevant items at the top of the recommendation list.
I will give a short overview of the ranking techniques that we developed the last couple of years and the main idea behind them. Recommendations should also be interesting and potentially allow users to discover new content and perhaps even expand his/her preferences.
In the second part of the talk, I will focus on Diversifying recommendations, the challenges and the ways we tackle them. I would also like to introduce a new Open Source project for Machine Learning and Recommendations in Giraph/Hadoop
called Okapi. Okapi provides a range of methods for Collaborative Filtering and Social Network Analysis and is released under the Apache licence.
Christoph Lingg – komoot – Recommender Use Case at komoot
komoot is your personal guide for cycling and hiking tours. Cristoph Lingg will give a short introduction about recommender use cases at komoot and their current recommendation techniques.