A Survey on Web Mining with Matrix Approximation

PhD Qualifying Examination


Title: "A Survey on Web Mining with Matrix Approximation"

Mr. Bin CAO


Abstract:

Many learning and data mining problems involve dyadic data that
can be represented by matrices. Matrix approximation models are im-
portant tools to explore the structure underlying these dyadic data.
In this survey, we give an overview on the matrix approximation mod-
els developed in the literature. By decomposing the models to several
components including loss function, constraints and regularization, we
give discussions on each component. Besides single matrix approxi-
mation models, we also survey the multiple matrices approximation
models including tensor-based approaches and collective matrix fac-
torization models. Furthermore, we discuss the applications of matrix
approximation models in Web mining related to social networks and
Web search.


Date:     		Friday, 15 May 2009

Time:                   10:00am-12:00noon

Venue:                  Room 5504
 			lifts 25-26

Committee Members:      Prof. Qiang Yang (Supervisor)
 			Dr. James Kwok (Chairperson)
 			Dr. Raymond Wong
 			Prof. Dit-Yan Yeung


**** ALL are Welcome ****