2014
EMSS 2014 Proceeding
Self-optimizing real-time recommendation model design based on YARN
Authors: Tao Liu, Shuang Wang, Peng Wu
Abstract
In real-time personalized recommendation systems, aimed at some existing issues such as timely system performance requirements,large data processing capacity,the existence of bidirectional cold-start,the difficulties of spotting the potential hot news,the deficiencies of system self-optimization and etc,we propose enhanced self-learning optimization model based on the YARN platform,which illustrates algorithm scheduling framework and self optimization system revised by users? feedback.Furthermore,to fully exhibit the superiority of the YARN platform,the label propagation algorithm here used exemplifies the polymerization process of raw large data.Thus,a highly prompt recommendation system overall is achieved.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
