Recommending the Best Value Hotels at Tripadvisor

Abstract: Tripadvisor is the world’s largest travel site, with an average of 340 unique monthly visitors and 500 million reviews and opinions. We provide recommendations for travellers using search log, review text, votes collected from our users, machine vision, and other machine learning techniques. This talk will talk about our recommendation system for hotel sorting, how we use Learning-to-Rank to decide the best value hotels for each user. This talk will also cover some interesting modelling issues that arise when building large numbers of models using terabytes of data and solutions we have developed.

Bio: Anyi is a machine learning scientist at Tripadvisor. She has a M.S. from Columbia University School of Engineering and Applied Science with a focus on Machine Learning. Anyi has lead multiple modeling projects across the company in search engine marketing, instant booking, photo identification, revenue optimization and sorting and recommendations."