Re-thinking Recommendation Systems in the Era of Conversational Interfaces

Abstract: Recommender systems have used machine-learning techniques extensively to improve the quality of many web properties. Many of these changes have occurred in the context of the existing user experience that is GUI-centric. Recent developments in natural language processing are enabling a new user experience that uses voice as an integral component. I will make the case that a voice-based recommender experience will be very different from the current recommender experience and will outline a research agenda around this coming change.  

Bio: Tushar Chandra is a Distinguished Engineer at Google, working in the intersection of machine learning and natural language processing. He was one of the creators of Sibyl, a large scale machine learning system that was widely used within Google. Prior to his work on Machine Learning, Tushar worked on large scale distributed systems such as Google's Bigtable and a fault-tolerant distributed consensus system that is widely used inside Google. Tushar received his Ph.D. in Computer Science from Cornell University in 1993 with Prof. Sam Toueg, then he worked at IBM Research, and was the lead architect at Tivoli software until he joined Google in 2004. He was a joint winner of the 2010 Edsger W. Dijkstra Prize in Distributed Computing, which was based on his work at Cornell.