A genetic algorithm for discriminative graph pattern mining
| Authors | |
|---|---|
| Year of publication | 2019 |
| Type | Article in Proceedings |
| Conference | Proceedings of the 23rd International Database Applications & Engineering Symposium, IDEAS 2019, Athens, Greece |
| MU Faculty or unit | |
| Citation | |
| Doi | http://dx.doi.org/10.1145/3331076.3331113 |
| Keywords | data mining; graph mining; dynamic graphs; pattern mining; discriminative patterns; random walk; genetic algorithm |
| Description | Real-world networks typically evolve through time, which means there are various events occurring, such as edge additions or at- tribute changes. We propose a new algorithm for mining discriminative patterns of events in such dynamic graphs. This is dierent from other approaches, which typically discriminate whole static graphs while we focus on subgraphs that represent local events. Three tools have been employed The algorithm uses random walks and a nested genetic algo- rithm to nd the patterns through inexact matching. Furthermore, it does not require the time to be discretized as other algorithms commonly do. We have evaluated the algorithm on real-world graph data like DBLP and Enron. We show that the method outperforms baseline algorithm for all data sets and that the increase of accuracy is quite high, between 2.5for NIPS vs. KDD from DBLP dataset and 30% for Enron dataset. We also discus possible extensions of the algorithm. |
| Related projects: |