Skip to main content

Machine Learning Applications: Recommendation Engines Using Multiple Behavior Sources

Ted Dunning (MapR Technologies)
Hardcore Data Science Gramercy Suite
Average rating: ****.
(4.25, 12 ratings)

Machine learning constructs such as Recommendation engines often take a simplistic approach to data modeling: a single kind of user interaction with a single kind of item is used to suggest the same kind of interaction with the same kind of item. In practice however, this approach is flawed for several reasons. First, multiple kinds of interactions with multiple kinds of items are typically available for training the recommendation engine to make suggestions. Second, recommendation is better viewed as a ranking problem rather than a regression problem. Finally, practical recommendation systems should be constantly self-training as today’s recommendations and selections can be used to train tomorrow’s recommender.

This session will shed light on a practical recommendation architecture and implementation style that addresses all of the above issues and which is considerably easier to implement and deploy than conventional approaches. Several of the techniques that I will describe have never (to my knowledge) appeared in the research literature. The session will also describe how the self-feeding and data-hungry nature of recommendation algorithms make supposedly secondary considerations like result order dithering more important than algorithm choice.

Photo of Ted Dunning

Ted Dunning

MapR Technologies

Ted Dunning has been involved with a number of startups—the latest is MapR Technologies, where he is chief application architect working on advanced Hadoop-related technologies. Ted is also a PMC member for the Apache Zookeeper and Mahout projects and contributed to the Mahout clustering, classification, and matrix decomposition algorithms. He was the chief architect behind the MusicMatch (now Yahoo Music) and Veoh recommendation systems and built fraud-detection systems for ID Analytics. Opinionated about software and data-mining and passionate about open source, he is an active participant of Hadoop and related communities and loves helping projects get going with new technologies.

Comments on this page are now closed.


10/29/2013 7:11pm EDT

They will be on the strata web site shortly. Also see

Picture of Benjamin Bengfort
Benjamin Bengfort
10/29/2013 11:33am EDT

To ask the question that commonly gets asked: is there a place we could get the slides?


Sponsorship Opportunities

For exhibition and sponsorship opportunities, contact Susan Stewart at

Media Partner Opportunities

For information on trade opportunities with O'Reilly conferences email mediapartners

Press & Media

For media-related inquiries, contact Maureen Jennings at

Contact Us

View a complete list of Strata + Hadoop World 2013 contacts