摘要
针对传统方法处理物联网中大数据时缺乏可扩展性,提出了基于MapReduce的物联网大数据处理框架,用于实现对各种形式的数据资源,包括结构化、半结构化和非结构化数据的处理,该框架为基于物联网的大数据智能应用提供了技术支持。使用K-最近邻技术来清除噪声数据;使用奇异值分解来降低数据的维数;使用模糊C均值和基于密度的空间聚类的混合技术来处理噪声数据。实验结果表明,所提出的框架具有较好的可扩展性和拟合精度,在IADL活动数据集上的准确率为98.9%。
Addressing the lack of scalability of traditional methods for handling big data in Internet of things,a big data processing framework of the Internet of things based on MapReduce is proposed to realize the processing of various forms of data resources,including structured,semi-structured and unstructured data.The framework provide technical support for the intelligent application of big data based on the Internet of things.K-nearest neighbor technique is used to remove noise data,and singular value decomposition is used to reduce the dimension of data.A hybrid technique of fuzzy C-means and density based spatial clustering is used to process noisy data.The experiment results show that the proposed framework has good scalability and fitting accuracy,and the accuracy on IADL activity data set is 98.9%.
作者
李红
邱凯
LI Hong;QIU Kai(Guizhou E-commerce Vocational and Technical College,Guiyang 551400,China)
出处
《信息技术》
2023年第7期173-178,共6页
Information Technology
关键词
物联网
MAPREDUCE
大数据
奇异值分解
KNN
Internet of things
MapReduce
big data
singular value decomposition
K-nearest neighbor
