演讲人: 微软亚洲研究院资深研究员 刘铁岩 博士

讲座时间: 2010年4月20日（周二）晚6：00~7：30

讲座地点: ダファベット 入金不要四层报告厅

讲座内容:

演讲人：微软亚洲研究院资深研究员 刘铁岩 博士

　　　　 Dr. Tie-Yan Liu，Lead Researcher（Microsoft Research Asia）

Dr. Tie-Yan Liu is a lead researcher at Microsoft Research Asia. His current research interests are theorem, algorithm and system for learning to rank in vector space and on graph data. So far, Dr. Liu has about 70 quality papers published in referred international conferences and journals, including SIGIR(9), WWW(3), ICML(3), KDD(2), etc. He has about 40 filed US / international patents or pending applications. He is the winner of the Most Cited Paper Award for the Journal of Visual Communication and Image Representation. He has been the program committee members for more than 30 international conferences, such as WWW, SIGIR, ICML, ACL, and ICIP. He has been a Senior Program Committee member of SIGIR 2008, and the co-chair of the SIGIR workshop on learning to rank for information retrieval in 2007 and 2008 (LR4IR 2007 and 2008). He has been on the Editorial Board of the journal of Information Retrieval. He has been a tutorial speaker at WWW 2008 and SIGIR 2008. Prior to joining Microsoft, Dr. Liu obtained his Ph.D. in electronic engineering from Tsinghua University in 2003, where his research efforts were devoted to video content analysis.

演讲摘要：

In the lecture, the ranking models for information retrieval will be introduced. First, we will talk about the conventional models such as Boolean model, Okapi model and language model. And then we will move onto the recent advances in using machine learning technologies for training a model. Some researchers proposed transforming ranking to classification and regression, and thus proposed using the pariwise approaches like Ranking SVM, RankBoost and RankNet to solve the problem of ranking. In recent years, however, some new methods were proposed based on deeper understanding of the ranking problem and information retrieval. Examples include ListNet, which defines a listwise loss function based on permutation probability distribution of ranked lists; AdaRank, which use Boosting techniques to optimize IR evaluation measures directly; and many others like SoftRank, SVM-MAP, etc. All these algorithms will be introduced in the lecture and after that we will discuss some future research directions in this field.