Mining Frequent Patterns with Counting Quantifiers

Yanxiao He, Xin Wang, Yuji Sha, Xueyan Zhong, Yu Fang

Abstract: Frequent Pattern Mining (FPM) is a classical graph mining task. In recent years, we have witnessed wide applications of FPM. However, emerging applications keep calling for more expressive patterns to capture more complex structures in a large graph. In light of this, we investigate the problem of mining frequent patterns with counting quantifiers. We first introduce quantified graph patterns (QGPs), which are the patterns incorporated with counting quantifiers. We then develop an algorithm along with an optimization technique to mine QGPs in a single large graph. On real-life graphs, we verified the performances of our (optimization) techniques.