The School of Informatics, Computing, and Engineering (SICE) Data Science Talk
Speaker: Xiaozhong Liu, Associate Professor of Information Science
Where: Luddy Hall, Rm. 1106, Dorsey Learning Hall
When: Friday, October 12, 2018, 10:00 am
Topic: Deep Natural Language Processing with 2.1 Billion Products + 0.5 Billion User Data
Abstract: In an active e-commerce environment, Natural Language Processing (NLP) can be critically important. In this talk, I will share the most recent NLP research works at Alibaba. By leveraging very large scale text and user behavior data, we proposed a number of novel algorithms to address different novel/classical research questions. Three cases will be presented: 1. Educate Machine to be a Creative Writer, 2. Sentiment Analysis with Text and User Behavior (Graph) Data, and 3. Cross-Domain Sentiment Transfer via Heterogeneous Graph Mining. In these works, we propose a number of novel deep learning algorithms by leveraging heterogeneous graph mining, DSSM-based RNN, and representation learning techniques.
Biography: Xiaozhong Liu is the Director of NLP Research, Alibaba DAMO Academy (also Associate Professor @ SICE/LIS). His research is focusing on Search, NLP, and Graph Mining. He is currently supervising Applied NLP and Text Mining team, and he proposed a number of innovative algorithms, including anti-spam classification, sentiment analysis, natural language generation (machine writing), dialog system and large-scale heterogeneous graph mining, to serve half billion users.