iMRAproiri: A Parallel Aproiri Algorithm Based on Twister

Abstract:

MapReduce is a current parallel computing model. Introduced MapReduce into Aproiri can make the algorithm more effective. But Apriori need to traverse the database several times to produce the final frequent itemsets accompanied with vast I/O consumption. Therefore, we put forward an algorithm named iMRAproiri by using Twister, an iterative MapReduce model, to reduce the I/O consumption. In iMRAproiri, the map function performs the procedure of counting potential candidates and the reduce function performs the procedure of summing the occurrences counts, while Twister handles the intermediate data in the distributed memory of the worker nodes. These features enable iMRAproiri to significantly reduce the amount of the time that I/O operation takes and increase the efficiency of Apriori algorithm accordingly.

• Periodical: Applied Mechanics and Materials (Volumes 599-601)
• Pages: 1863-1867
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.599-601.1863
• Online since: August 2014
• Authors: Liang Huang*, Zhi Jing Liu
• Keywords: Aproiri, MapReduce, Twister

References

[1] Shan Huang, Botao Wang, Guoren Wang, et al: Journal of Frontiers of Computer Science and Technology Vol. 7 (2013), pp.865-885.

[2] Dean J, Ghemawat S: MapReduce: Communications of the ACM Vol. 51 (2008), pp.107-113.

[3] Ekanayake J, Li H, Zhang B, et al: Proceedings of the 19th ACM International Symposium on High Performance Distributed Computing (2010), pp.810-818.

[4] Wu X, Kumar V, Quinlan J R, et al: Knowledge and Information Systems Vol. 14 (2008), pp.1-37.

[5] Junqiang Liu, Xiaoying Sun, Yunhe Pan: Journal of Computer Science Vol. 31 (2004), pp.110-113.

[6] Meifeng Zhang, Jianwei Zhang: Journal of Computer engineering and application Vol. 39 (2003), pp.196-198.

[7] Information on http: /www. iterativemapreduce. org.