分布式集群环境下基于并行计算的图聚类信息高效处理方案  

Efficient Graph Clustering Information Processing Scheme Based on Distributed Cluster Environment

作　　者：张博[1] 

机构地区：[1]长沙师范学院电子与信息工程系,湖南长沙410100

出　　处：《自动化技术与应用》2016年第3期29-35,67,共8页Techniques of Automation and Applications

基　　金：湖南省自然科学基金项目(2015JJ6007);湖南省教育厅科学研究项目(13C1070)

摘　　要：针对人工智能领域图聚类数据分析与处理能力无法适应于日益复杂的分布式集群环境等问题,提出一种基于并行计算的高效率图聚类信息处理方案。在分布式集群计算环境下对超大规模、超大区域范围图数据信息的稀疏化分析与处理机制上,通过对Minhash算法以Map Reduce架构理论进行改进,使其实现对数据的并行化分析处理,确保能够在日益复杂的分布式集群计算环境下高效处理图聚类数据信息。实验表明,改进方案不仅可行,而且能够对图聚类数据信息进行快速稀疏化处理,具有一定的高效性。In order to the problem about graph clustering data analysis ability can＇t adapt to the increasingly complex distributed cluster environment in artificial intelligence, this scheme designes an efficient sparsification algorithm based on parallel computing. This algorithm analyzes and improves Minhash algorithm based on Map Reduce framework theory, and efficiently processes and analyzes the graph clustering data information in the increasingly complex distributed cluster environment. Experiment showes that the algorithm not only is feasible, but also can quickly process graph clustering sparse data, and it has a high efficiency.

关 键 词：人工智能 数据挖掘 映射规约 图聚类 最小哈希算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]