Flexible Support Measures and Search Schemes for Mining Frequent Patterns in a Large Single Graph
Jinghan Meng · 2017
Recent years have witnessed intensive studies on mining graph databases for interesting patterns. One important problem in this paradigm is the Frequent Subgraph Mining (FSM), which involves finding patterns that occur in high frequency from a graph database. In particular, researchers are interested in finding all frequent patterns from a single graph that often consists of a large number of vertices and edges. There are two indispensable components in any solutions to the FSM problem. The first one is a support measure that quantifies the frequency of a given pattern occurred in the data graph. The second one is an algorithm for searching through the space of all possible subgraphs to identify frequent ones. These two computations, if handled in a naive way, face serious performance and accuracy issues. The main objective of my research project is to study both processes by providing theoretical breakthroughs and developing practical algorithms to further speed up the processing of FSM.