Name
Abstract | ||
|---|---|---|
Mining frequent itemsets is an essential data mining problem. As the big data era comes, the size of databases is becoming so large that traditional algorithms will not scale well. An approach to the issue is to parallelize the mining algorithm, which however is a challenge that has not been well addressed yet. In this paper, we propose a MapReduce-based algorithm, Peclat, that parallelizes the vertical mining algorithm, Eclat, with three improvements. First, Peclat proposes a hybrid vertical data format to represent the data, which saves both space and time in the mining process. Second, Peclat adopts the pruning technique from the Apriori algorithm to improve efficiency of breadth-first search. Third, Peclat employs an ordering of itemsets that helps balancing the workloads. Extensive experiments demonstrate that Peclat outperforms the existing MapReduce-based algorithms significantly. |
Year | DOI | Venue |
|---|---|---|
2015 | 10.1007/978-3-319-22849-5_27 | DEXA |
Field | DocType | Citations |
Data mining,Data format,Computer science,Parallel algorithm,Apriori algorithm,FSA-Red Algorithm,Data mining algorithm,Big data,Database | Conference | 1 |
PageRank | References | Authors |
0.35 | 14 | 6 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jun-Qiang Liu | 1 | 62 | 5.64 |
| Yongsheng Wu | 2 | 2 | 2.38 |
| Qingfeng Zhou | 3 | 1 | 0.68 |
| Benjamin C. M. Fung | 4 | 2062 | 90.87 |
| Fanghui Chen | 5 | 1 | 0.35 |
| Binxiao Yu | 6 | 1 | 0.35 |