食品安全大数据的融合及分类并行处理技术研究  

Research on Fusion and Classification Parallel Processing Technology of Food Safety Big Data

作　　者：张素智[1] 杨芮[1] 赵亚楠[1] ZHANG Suzhi;YANG Rui;ZHAO Yanan(School of Computer and Communication Engineering,Zhengzhou University of Light Industry,Computer Applications and Software,Zhengzhou 450002,China)

机构地区：[1]郑州轻工业学院计算机与通信工程学院,郑州450002

出　　处：《湖北民族学院学报（自然科学版）》2018年第3期256-265,共10页Journal of Hubei Minzu University(Natural Science Edition)

基　　金：国家自然科学基金项目(61672470);北京市重点实验室开放课题(BKBD-2017KF08)

摘　　要：食品安全大数据具有多源、高层次、强关联等特征,通过对食品安全大数据挖掘处理可快速高效地发掘数据的潜在价值,帮助提高食品安全态势感知及预测、病因性食品关联等综合分析能力.对食品安全大数据的融合及分类并行处理技术进行综述.介绍了食品安全大数据的来源、类型和特征并总结其关键处理技术;阐述了食品安全大数据预处理方法即数据融合技术;归纳了食品安全大数据挖掘技术,包括并行处理的三种计算模型以及多种聚类方法,如减法聚类、K-Means经典聚类、核聚类及谱聚类等;最后,对食品安全大数据未来的挑战和研究方向进行总结和展望.The big data of food safety has such characteristics as multi-source,high-level and strong correlation.By mining and processing food safety data,the potential value of data can be quickly and efficiently explored,which help to improve the comprehensive analysis ability of food safety situation awareness,prediction and etiological food association. This paper reviews the fusion and classification parallel processing technology of food safety big data.Firstly,this article introduces the source,types and characteristics of food safety big data and summarizes its key processing technologies. Secondly,the data fusion technique of big data food safety preprocessing is described.Then,the paper concludes the food safety big data mining technology,including three computing models for parallel processing and multiple clustering methods,such as subtractive clustering,K-Means classical clustering,nuclear clustering and spectral clustering,and so on. Finally,the future challenges and research direction of big data of food safety are discussed.

关 键 词：食品安全大数据 数据融合 并行处理 数据挖掘 聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]