Who: Prof. Xinhua Zhuang (C.W. LaPierre Chair Professor, University of Missouri, USA)
When: Thursday, May 12, 2011. 10:00 PM.
Approximate duration of 2 hour(s).
Where: Room 314, Tower B, Fundamental Teaching Building, Wangjiang Campus
What: In this lecture, Prof. Zhuang takes a fundamentally different approach for Web page ranking. He first propose to establish pair-wise similarity measures among pages returned by text-based search engine by combining local link structure information with other features, which are independent of link structure yet contributable to page rank, and justify the use of principle eigenvector of similarity matrix for the importance based page ranking. He then apply densely connected clustering for topic analysis and construct the final page ranking vector by combining importance based page ranking with topic diversity. The proposed approach represents a generic framework in Web page ranking. It allows to integrate more ranking contributable features and to fully exploit similarity measures in delivering more accurate page ranking with topic diversity.