Weiran Xu
Associate Professor, Master Supervisor, Ph.D Supervisor
Beijing University of Posts and Telecommunications
His research focuses on solving text processing problems based on machine learning and pattern recognition, such as text classification, information retrieval, information extraction and propensity judgment. Since 1997, he has been engaged in the field of pattern recognition and machine learning, and since 2003, he has been specialized in the research of text data machine learning. He participated in TREC, TAC, 863, COAE and other related evaluations since 2004, and won the first place in individual and comprehensive scores for many times. He was responsible for building a series of prototype systems; he has participated in a number of National Natural Science Foundation of China, 863 projects and major national science and technology projects as a major member. He has published nearly 10 papers including ACL, AAAI, SIGIR and other top conference papers, more than 10 SCI indexed journal papers, more than 50 EI indexed papers.
The network contains all kinds of useful information, and the bottleneck is how to get it automatically. His long-term research has been to enable computers to automatically understand the content of text and provide services to people voluntarily. There is still a lot of room for improvement in the capabilities of machines compared to the capabilities of people. But it is still very difficult for machines to comprehensively surpass people in the present situation.
At present, the main research problem is to organize the content of text with entity or event as the center, so as to solve the problems of information extraction, information retrieval, text classification and tendency judgment. The main theories and methods adopted are representation learning theory and complex network theory. Deep Learning in Representation Learning (or Feature Learning) has achieved excellent results in image and speech processing. The theory of representation learning is still in the preliminary stage of research, and the commonly used methods mainly include “probability model”, “automatic coding” and “manifold learning”. The dynamic model proposed by Professor Jun Guo in our lab based on complex network has good effect on mining and representing various factors and their correlation relations. Therefore, applying it in the framework of presentation learning theory will better solve the problem of text content extraction and representation.
Interests
- Information Retrieval
- Pattern Recognition
- Machine Learning
Education
-
Associate Professor, Master Supervisor, 2006.7
Beijing University of Posts and Telecommunications
-
Lecturer, 2003.7-2006.7
Beijing University of Posts and Telecommunications
-
Signal and Information Processing Ph.D, 2003.7
Beijing University of Posts and Telecommunications
-
Signal and Information Processing Master's Degree, 2000.6
Dalian University of Technology
-
Information Engineering Bachelor's Degree, 1997.7
Dalian University of Technology