Main Article Content
Abstract
Huge volume of data values are collected and managed under the big data models. The big data mining operations are performed with the support of the storage and computational resources from the clouds. The big data analytics are carried out using the Hadoop environment with parallel and distributed computing support. The homogeneous and heterogeneous computing models are managed with the Hadoop environment. The data and tasks are partitioned and processed as small elements in the MapReduce framework. The mapreduce framework is adapted to manage the data and task as tiny elements. The association rule mining methods are applied to detect the frequent patterns in the data values. The data values are equally divided and processed under the parallel frequent pattern mining techniques. The rule mining with load management operations are carried out with the Data Partitioning in Frequent Itemset Mining on Hadoop Clusters (FiDoop-DP). The transaction relationships are analyzed in the Voronoi diagram based data partitioning scheme. The similarity metric and Locality Sensitive Hashing (LSH) technique are applied to manage the redundant data values. The Parallel Frequent Pattern Growth algorithm is employed to discover the rules using the cloud resources. The data pattern discovery operations are designed with resource and data constraints.
The dynamic data partitioning and transmission over the Hadoop clusters is supported in the parallel rule mining process. The dynamic resource level based computational nodes are used to construct the heterogeneous Hadoop clusters. The load management in data partitioning operations are carried out with data and resource constraints. The FiDoop-DP scheme is enhanced to handle the data placement with load management under the Hadoop Distributed File System (HDFS) in heterogeneous nodes. The computational and communication loads are minimized with the parallel frequent pattern mining process.