Efficiently Mining Closed Frequent Patterns with Weight Constraint from Directed Graph Traversals Using Weighted FP-Tree Approach

Abstract

In this paper, a transformable model of EWDG (edge-weighted directed graph) and VWDG (vertex-weighted directed graph) is proposed to resolve the problem of weighted traversal patterns mining. Based on the model, an effective algorithm called GTCWFP miner (graph traversals-based closed weighted frequent patterns miner) is presented. The algorithm exploits a divide-and-conquer paradigm with a pattern growth method to mine closed frequent patterns with weight constraint from the traversals on directed graph. It incorporates the closure property with weight constrains to reduce effectively search space and extracts succinct and lossless patterns from graph traversal TDB. Experimental results of synthetic data show that the algorithm is an efficient and scalable algorithm for mining closed weighted frequent patterns based on graph traversals.