Machine Learning Data Scientist
Liang Wu is a machine learning data scientist at Airbnb, currently focusing on building search and recommender systems of Airbnb Experiences. He has published over 30 papers in major AI conferences including AAAI, ICDM, ICWSM, SIGIR, and WSDM, and he serves as Program Committee member of AAAI, SIGIR, KDD, WSDM, BIGDATA, etc. Liang was the main speaker for two tutorials in SBP'16 and ICDM'17. In addition, Liang authored 2 patents, 2 book chapters, and is the first author for the incoming book on Misinformation and Spam Detection in Social Media, from Cambridge University Press. He obtained Ph.D. with Professor Huan Liu from Arizona State University.
Liang Wu is speaking at the following session/s
Tackling a “Small Data” Search Challenge at Airbnb Experiences
Learn about two projects in Airbnb Experiences, a new service of Airbnb, highlighting the use of causal inference in an ecommerce search engine. A new service usually faces challenges brought by small data: inadequate data and noisy label information, which make it difficult to utilize user search logs. This problem can be easily solved when the user traffic is large enough to support large-scale randomized experiments. For an early-stage product, however, it is particularly difficult, if not impossible. In this talk, learn how the Data Science team of Airbnb Experiences solve this problem with causal inference. Explore two natural experiments designed to make biased feedback directly usable without running randomized experiments.
Attendees will understand how to approach ecommerce search for an early-stage product, with a corresponding small data challenge. Learn the concept of causal inference and how to design natural experiments to remove biases in search log data.
Practitioners of ecommerce search and researchers interested in using user-generated feedback data. Basic knowledge about the workflow of a search engine will be needed.