We validate the discovered clusters through carefully investigating the lead authors’ affiliations from their homepages. However, identifying a leader in a link based cluster is a well known ranking problem, for which there are large volume of measurements. Here, we just use the node’s internal degree 3 to evaluate the importance of a node in a cluster. The larger the internal degree, the more likely the node is a leader.Figure 9 shows four main clusters discovered at the end of 2008 marked by some lead members. Through carefully checking the homepages of the authors, Figure 9a corresponds to Information Retrieval and Mining Group of Microsoft Research Asia, where Tie-yan Li and Hang Li are the lead researcher and senior researcher respectively. The lead authors in Fig. 9b are from University of Illinois at Urbana-Champaign (Jiawei Han) and IBM T.J.Watson (FangWei), which suggests there exist collaboration between the two organizations. The leaders in Fig. 9c are from Yahoo! research and Carnegie Mellon University, which implies that the two research groups have some collaborations. Ravi Kumar is an active researcher in Yahoo! research and Christos Faloutsos is a professor of Carnegie Mellon University, who supervised Jure Leskovec. Figure 9d is a data mining research group from Arizona State University, where Huan Liu is a professor and Jianping Ye is an associate professor and Zheng Zhao is co-supervised by the above two authors.
đang được dịch, vui lòng đợi..
