oktech Data Sense # 3: Recommender Systems

On November 6, at the Odnoklassniki company’s Moscow office, approx. Tech Data Talk # 3 will be held, this time we decided to dedicate the event to recommendation systems. Together with colleagues from OK.ru, Joom and SkolTech, we will talk about the past RecSys19, as well as about the theory, practice and trends of recommendation systems. Vlad Grozin will review the iconic RecSys19 conference. Eugene Frolov will tell one of the reports on HybridSVD, which is used to build hybrid recommendation systems. Then we move from theory to practice, and Andrey Kuznetsov will share his practical experience in improving recommendation systems for Odnoklassniki groups. As always, after the presentations there will be a discussion where everyone can ask any question to the speakers. The event will be hosted by Alexey Chernobrov.



We are waiting for everyone who is interested in the topic of creating, improving and operating recommender systems.



Register for the event.

To be or not to be RecSys 2019 - Vladislav Grozin, Joom

Many people want to go to some kind of conference devoted to big data analysis or even better recommendation systems, however, this is a serious investment of both time and effort, and company funds, of course, if you are not rich enough to pay for everything yourself :)

But how is it worth it?

In my report, I will tell you about my trip to RecSys19, we will understand that it was interesting, what you can get from participating in such an event (except for stickers on a laptop), and what will fail. Hope this helps dispel doubts and some myths.

But that's not all! I will talk about some of the most interesting, in my opinion, pagers that were presented at the conference.

HybridSVD pros and cons - Evgeny Frolov, SkolTech

This talk will be about the recently introduced HybridSVD model for building hybrid recommendation systems. We will discuss the key features of the model, its differences from existing matrix factorization methods and its main advantages. The purpose of the report is to show the simplicity and effectiveness of the proposed approach in terms of practical application.

How we make group recommendations in OK - Andrey Kuznetsov, OK.ru

Conventional referral systems are key user companions that truly make life easier in a world of ever-growing space of choice. In the report, I will talk about how we use OK to improve group recommendations and how we fit into the infrastructure of a large high-load project with tens of millions of users with such solutions.

timetable

19:00 - 19:10 Opening remarks, Alexey Chernobrov

19:10 - 19:30 To be or not to be RecSys 2019, Vladislav Grozin, Joom

19:30 - 19:50 HybridSVD pros and cons, Evgeny Frolov, SkolTech

19:50 - 20:10 How we make group recommendations in OK, Andrey Kuznetsov, OK.ru

20:10 - 20:30 Break

20:30 - 21:30 Discussion on the present and future of recommendation systems



Register for the event.