Presented By O’Reilly and Cloudera
Make Data Work
September 11, 2018: Training & Tutorials
September 12–13, 2018: Keynotes & Sessions
New York, NY

Schedule: Recommendation Systems sessions

Machine learning has been used extensively for recommendation and personalization applications. The rise of new algorithms tends to bring renewed interest. Deep learning has caused many companies to evaluate their existing recommenders, and many have begun to use neural networks to either supplement or replace their existing models.

9:00am–12:30pm Tuesday, 09/11/2018
Deep learning-based search and recommendation systems using TensorFlow
Location: 1E 15/16 Level: Intermediate
Vijay Agneeswaran (Walmart Labs), Abhishek Kumar (Publicis Sapient)
Average rating: ****.
(4.40, 5 ratings)
Abhishek Kumar and Vijay Srinivas Agneeswaran offer an introduction to deep learning-based recommendation and learning-to-rank systems using TensorFlow. You'll learn how to build a recommender system based on intent prediction using deep learning that is based on a real-world implementation for an ecommerce client. Read more.
9:00am–5:00pm Tuesday, 09/11/2018
Findata Day
Location: 1A 08
Alistair Croll (Solve For Interesting), Robert Passarella (Alpha Features), Amro Alkhatib (National Health Insurance Company-Daman), Mridul Mishra (Fidelity Investments), Patrick Angeles (Cloudera), James Psota (Panjiva ), Andreas Kohlmaier (Munich Re), Paul Lashmet (Arcadia Data), Nick Curcuru (Mastercard), Robin Way (Corios), Theresa Johnson (Airbnb), Jane Tran (Unqork), Swatee Singh (American Express)
From analyzing risk and detecting fraud to predicting payments and improving customer experience, take a deep dive into the ways data technologies are transforming the financial industry. Read more.
11:20am–12:00pm Wednesday, 09/12/2018
Semantic recommendations
Location: 1A 06/07 Level: Beginner
Shioulin Sam (Cloudera Fast Forward Labs)
Average rating: ***..
(3.25, 4 ratings)
Recent advances in deep learning allow us to use the semantic content of items in recommendation systems, addressing a weakness of traditional methods. Shioulin Sam explores the limitations of classical approaches and explains how using the content of items can help solve common recommendation pitfalls, such as the cold start problem, and open up new product possibilities. Read more.
1:15pm–1:55pm Wednesday, 09/12/2018
Harnessing and customizing state-of-the-art recommendation solutions with OpenRec
Location: 1A 15/16 Level: Intermediate
Longqi Yang (Cornell Tech, Cornell University)
State-of-the-art recommendation algorithms are increasingly complex and no longer one size fits all. Current monolithic development practice poses significant challenges to rapid, iterative, and systematic, experimentation. Longqi Yang explains how to use OpenRec to easily customize state-of-the-art solutions for diverse scenarios. Read more.
1:15pm–1:55pm Wednesday, 09/12/2018
Document vectors in the wild: Building a content recommendation system for Reuters.com
Location: 1A 06/07 Level: Intermediate
James Dreiss (Reuters)
Average rating: ***..
(3.67, 3 ratings)
James Dreiss discusses the challenges in building a content recommendation system for one of the largest news sites in the world, Reuters.com. The particularities of the system include developing a scrolling newsfeed and the use of document vectors for semantic representation of content. Read more.
2:05pm–2:45pm Wednesday, 09/12/2018
Achieving personalization with LSTMs
Location: 1A 15/16 Level: Intermediate
Ankit Jain (Uber)
Average rating: ***..
(3.00, 3 ratings)
Personalization is a common theme in social networks and ecommerce businesses. Personalization at Uber involves an understanding of how each driver and rider is expected to behave on the platform. Ankit Jain explains how Uber employs deep learning using LSTMs and its huge database to understand and predict the behavior of each and every user on the platform. Read more.
2:05pm–2:45pm Wednesday, 09/12/2018
Diversification in recommender systems: Using topical variety to increase user satisfaction
Location: 1A 06/07 Level: Intermediate
Ahsan Ashraf (Pinterest)
Online recommender systems often rely heavily on user engagement features. This can cause a bias toward exploitation over exploration, overoptimizing on users' interests. Content diversification is important for user satisfaction, but measuring and evaluating impact is challenging. Ahsan Ashraf outlines techniques used at Pinterest that drove ~2–3% impression gains and a ~1% time-spent gain. Read more.
2:55pm–3:35pm Wednesday, 09/12/2018
Perverse incentives in metrics: Inequality in the like economy
Location: 1A 06/07 Level: Intermediate
Bonnie Barrilleaux (LinkedIn)
Average rating: ****.
(4.50, 4 ratings)
As LinkedIn encouraged members to join conversations, it found itself in danger of creating a "rich get richer" economy in which a few creators got an increasing share of all feedback. Bonnie Barrilleaux explains why you must regularly reevaluate metrics to avoid perverse incentives—situations where efforts to increase the metric cause unintended negative side effects. Read more.
4:20pm–5:00pm Thursday, 09/13/2018
Building turnkey recommendations for 5% of internet video
Location: 1A 21/22 Level: Intermediate
Nir Yungster (JW Player), Kamil Sindi (JW Player)
JW Player—the world’s largest network-independent video platform, representing 5% of global internet video—provides on-demand recommendations as a service to thousands of media publishers. Nir Yungster and Kamil Sindi explain how the company is systematically improving model performance while navigating the many engineering challenges and unique needs of the diverse publishers it serves. Read more.