摘要
大数据往往呈现出大规模性、多模态性以及快速增长性等特征.粒计算是智能信息处理领域中大规模复杂问题求解的有效范式.从推动大数据挖掘研究角度,本文首先概要地讨论了大数据的特征对可计算性、有效性与高效性提出的3大挑战;其次,结合粒计算的思维模式特点,概述了已有研究成果,分析论述了以粒计算应对大数据挖掘挑战的可行性,认为粒计算有望为大数据挖掘提供一条极具前途的崭新途径;最后,对大数据挖掘的粒计算理论与方法中的若干科学问题进行了梳理与展望,以期抛引这一领域的学术思考与研究工作.
The external torm of big data often presents large-scale, multiple Inodal, and growth characteristics. In this paper, we discuss and analyze the ehallenges in data mining from the viewpoint of big data; these challenges include computability, effectiveness, and efficiency. Granular computing is an effective method for solving complex problems for intelligent information processing. By analyzing the feasibility of large data analysis based oil granular computing, we argue that gramular computing shows great promise as a new way for data mining in the context of big data. We also analyze several important problems in data mining based on granular computing, and the results will lead to further interpretations and developments in the field of big data mining.
出处
《中国科学：信息科学》
CSCD
北大核心
2015年第11期1355-1369,共15页
Scientia Sinica(Informationis)
基金
国家自然科学基金(批准号:61432011,U1435212,61322211)
国家重点基础研究发展计划(973计划)(批准号:2013CB329404)
教育部新世纪人才支持计划(批准号:NCET-12-1031)资助项目
关键词
大数据
数据挖掘
模式发现
粒计算
信息粒化
多粒度
big data, data mining, pattern discovery, granular computing, information granulation, multigran-ulation
