Frequent graph mining has been studied extensively with many scalable graph mining algorithms developed in the past. Graph patterns are essential not only for exploratory graph mining but also for advanced graph analysis tasks such as graph indexing, graph clustering, and graph classification. In this chapter, we examine the frequent pattern-based classification of graph data. We will introduce different types of patterns used in graph classification and their efficient mining approaches. These approaches could directly mine the most discriminative subgraphs without enumerating the complete set of frequent graph patterns. The application of graph classification into chemical compound analysis and software behavior prediction will be discussed to demonstrate the power of discriminative subgraphs.
|Original language||English (US)|
|Title of host publication||Link Mining|
|Subtitle of host publication||Models, Algorithms, and Applications|
|Publisher||Springer New York|
|Number of pages||26|
|State||Published - Jan 1 2010|
ASJC Scopus subject areas