[Pycon] [new paper] "Riccardo Lorenzon" - A Reinforcement Learning powered Recommender Engine

[info a pycon.it]

Title: A Reinforcement Learning powered Recommender Engine
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: it
Type: Talk

Abstract: Assessing the quality of a recommender system is a tricky job. 
Providing good recommendations does not mean only being able to understand user content preferences but mostly assuring the right mix of accuracy, novelty and diversity. A common approach is A/B testing, yet we prefer online solutions based on a simple reinforcement learning approach when the number of users and the number of interactions with the system are limited. 
Our goals are to share all the practical lessons we learned during the development of a book recommender and unveil the mix of tools and algorithms that proved to be effective for our case. 
The ideal attendee is a Basic level Python Developer for Data Science, with a brief knowledge of Hadoop framework and NoSQL databases.

Tags: [u'mongodb', u'REST', u'Python', u'A/B-testing', u'reinforcement-learning', u'recommender-systems', u'machine-learning', u'spark', u'hadoop', u'open-source', u'data-science']