Large-Scale Social Recommender Systems at LinkedIn
Recommender Systems are very important for social networks such as LinkedIn for engaging users, discovering new content, connections, opportunities, and satisfying users needs. This talk gives an overview of various analytics products at LinkedIn such as Jobs recommendations, Skills endorsements, People You May Know, and LinkedIn Today. Each of these products can be viewed as a large scale social recommendation problem, which analyzes billions of possible options, and suggests appropriate recommendations.
This talk will also touch upon various big data technologies such as Hadoop MapReduce, Azkaban work-flow management, Voldemort key-value store, and Kafka publish-subscribe system, to process terabytes of data for building large-scale social recommendation systems. Overall, people interested in building interesting social data products using machine learning and big data technologies will greatly benefit from this talk.