Today’s recommendation systems ingest a wide range of information, such as diverse user feedback signals (ratings, clickthrough, likes, and views) and auxiliary, contextual, and cross-platform traces (images, video, audio, and other associated metadata, as well as social networks and personal digital traces). A state-of-the-art system usually involves numerous heterogeneous and complex submodels that analyze and fuse high-dimensional and multichannel data streams.
Current development practice usually treats a recommendation algorithm as singular and monolithic. As a result, in order to experiment with a new method for even a small part of an algorithm or customize an algorithm for other application scenarios, researchers and practitioners need to reimplement the whole model from scratch or extensively patch existing code.
OpenRec, an open source framework that modularizes recommendation algorithms, was designed to tackle these challenges. Each recommender is modeled as a structured ensemble of reusable modules with standard interfaces. Under such a paradigm, changes to a module or the computational graph do not affect other components, and development and testing can be more readily achieved via plug-ins.
Longqi Yang explains how to use OpenRec to easily customize state-of-the-art solutions for diverse scenarios.
Topics include:
Longqi Yang is a PhD candidate in computer science at Cornell Tech and Cornell University, where he is advised by Deborah Estrin, and is a member of the Connected Experiences Lab and the Small Data Lab. His current research focuses are user modeling, recommendation systems, and recommendation for social good. His work has been published and presented in top academic conferences, such as WWW, WSDM, Recsys, and CIKM. He co-organized workshops at the NYC Media Lab annual summit 2017 and KDD 2018.
For exhibition and sponsorship opportunities, email strataconf@oreilly.com
For information on trade opportunities with O'Reilly conferences, email partners@oreilly.com
View a complete list of Strata Data Conference contacts
©2018, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • confreg@oreilly.com