A Distributed Partitioning Algorithm for E-Government Service Platform based on Cloud Computing
Download as PDF
DOI: 10.25236/meici.2019.106
Author(s)
Yonglin Leng, and Xiaohong Sun
Corresponding Author
Yonglin Leng
Abstract
Network public opinion from the perspective of E-government Crisis management can effectively govern the virtual society. The monitoring and prediction of public opinion can not be separated from the data. The single-machine model is inefficient for the large number of public opinion data. The emergence of distributed technology based on cloud computing improves the processing efficiency of public opinion. But a key technology of distributed processing is the data partitioning. In order to mine the public opinion data more deeply, RDF (Resource Description Framework) is used to describe network resource, which can achieve more efficient and accurate retrieval. This paper will study how to partition the network public opinion data based on RDF. The star fabric existing extensively is an important structure of RDF data. Aiming at the star fabric, we propose a distributed partitioning algorithm based on star-based. First, the 2 hops star fabric is built with Hadoop. And next a weighted graph is constructed based all these star fabrics, and a balance K-medoids clustering algorithm is adopted to divided the weighted graph. Experimental result gets a lower replication ratio and a better load balancing.
Keywords
E-government, Network public opinion, RDF, Partitioning