An improved pattern mining technique for graph pattern analysis using a novel behavior of artificial bee colony algorithm

Abstract

Rising data complexity and volume in the network has attracted researchers towards substructure analysis. Subgraph mining is an area that has gained remarkable attention in the last couple of years to offer an intelligent analysis of more massive graphs and complicated data structures. It has been observed that graph pattern mining faces issues regarding the matching ruleset and complex instruction set execution problem. This paper introduces modern-day intelligence architecture based on Swarm Intelligence that is cross-validated by supervised machine learning mechanisms. A new behavior incorporated with a new inter and intra hive behavior is incorporated in Swarm based Artificial Bee Colony. The proposed work model is evaluated over two different datasets with more than 4900 nodes in the graph. The proposed framework is evaluated using True Detection Rate, False Detection Rate, precision, and F-Measure, demonstrating an average improvement of 9.8%, 8.35%, 8.35% and 9.15% against existing GraMi work that represent an enhanced performance of the proposed pattern mining technique.