ABSTRACT
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Link Analysis deals with mining useful information from linked structures like graphs. Graphs have vertices representing objects and links among those vertices representing relationships among those objects. Data mining covers a diverse set of activities that deal with independent sets of data that may be numerical or otherwise. Link mining works with graph structures that have nodes with defined set of properties. These nodes may be of the same type (homogeneous) or different (heterogeneous). The World Wide Web can be considered a homogeneous network (all nodes are web URLs), while a network of bibliographic links and citations is a heterogeneous multi-mode network (nodes may be papers, books, magazine articles, etc.) [15]. We first take a very generic look at links, and establish what kind of information and benefits can be obtained by studying links.
