项目名称: 基于相关性的大数据分类理论与方法研究
项目编号: No.71471060
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 管理科学
项目作者: 陈德刚
作者单位: 华北电力大学
项目金额: 62万元
中文摘要: 在大数据决策问题的背景之下,本项目以融合大数据决策问题的基本特点为出发点,以大数据中存在相关性为基本假设前提,以基于全部数据挖掘大数据中隐含的相关关系为基本目标,通过引入相关测度、相关算子、相关决策系统和相关学习等基本概念建立大数据分类的基本数学模型,在此基础之上提出相关学习一致性的概念和利用相关算子进行预测的新模式,建立求解最优相关算子的优化模型并针对不同类型的数据设计各自的求解算法,针对大数据的容量特点开发各种大数据分类的算法并进行实验验证,研究这些算法的并行式和增量式的实现方式,并利用实验分析与比较来验证所提出的理论与方法的有效性.本项目的完成将会建立基于相关性的大数据分类基本理论框架并实现对大数据决策问题更有效的分析.
中文关键词: 数据挖掘;知识发现;大数据;信息系统
英文摘要: This proposal focuses on big data decision problems by considering fusion of basiccharacters of big data decision problems by big data as starting point.With assumption of existence of correlativity among big data, this proposal aims to reveal correlation hiding in big data in terms of the population of the whole data set.The fundamental mathematical model of big data classification is set up by developing some notions such as correlative measure, correlative operator,correlative decision system and correlative learning. Based on this model, notion of consistance of correlative learning and a new forecasting model with correlative operator are proposed, while an optimization model to capture the optimal correlative operator is set up and different methods are developed for different type of data. Furthermore, several algorithms for big data classification are put forward to address the huge volume of big data and tested with some experiments. The effectiveness of the theory and methods in this proposal will be demostrated by performing experiments. Results in this paper will contribute to setting up fundamental framework for big data classification in terms of correlativity and effective analysis of big data decision.
英文关键词: data mining;knowledge discovery;big data;information systems