MapReduce-Based H-Mine Algorithm

Frequent Item set Mining (FIM) is a very effective method for knowledge acquisition from data, but with the advent of the era of big data, traditional algorithms based on memory are facing severe challenges such as the computation speed and storage capacity. Fortunately, Map Reduce model provides an efficient framework for distributed programming and operation framework. This paper proposes a novel Map Reduce-based H-mine algorithm (MRH-mine), a version of H-mine algorithm in the distributed operation environment. Experimental results show that MRH-mine algorithm has a better performance and scalability than traditional H-Mine when facing massive data growth.