Great and opinionated summary of RecSys 2016, and also somewhat of a summary of recommendations and personalization research as of 2016. The big take aways for me were that deep learning is just starting and so far unproven, and the research community is finally turning away from RMSE of explicit ratings for evaluation (which is relatively useless in practice; most real problems need TopN recommendations from implicit data).
http://blog.gravityrd.com/recsys-2016-research-perspective-part/
Shared publiclyView activity