How to win friends and influence people (with recommendations)

Jueves 26 - 17:20

Sam Shah, LinkedIn

Summary: Recommendations are an integral part of any web property for discovery and engagement. At LinkedIn, the largest professional social network with 160+ million users, we have over 30 different recommendation products. Some examples include "People You May Know", a link prediction system; "Groups You May Like", a group discovery mechanism; "Viewers of this Profile Also Viewed", a collaborative filtering application; and "Related Searches", a search query refinement and exploration tool; among others.

Such practical problem solving with data involves more than just applying the latest machine learning techniques: intuition, domain knowledge, and reasonable approximations can mean the difference between a successful model and a catastrophic failure. In this talk, I will deep dive into the construction of these recommendation systems, providing flavor into real-life data mining challenges.

Bio : Sam Shah is a principal engineer on the LinkedIn data team. He leads many of the site's large-scale recommendation and analytics systems, which analyze hundreds of terabytes of data daily to produce products and insights that serve LinkedIn's members. His work involves pure research, product-focused features, and infrastructure development, including social network analysis, recommendation engines, distributed systems, and grid computing. Some of the products under his purview include "People You May Know", "Who's Viewed My Profile?", Skills & Expertise, related searches, and more. Sam holds a Ph.D. in Computer Science from the University of Michigan.