干预规则挖掘的概念、任务与研究进展  

Concepts,Tasks and Research Advances of Intervention Rule Mining

作　　者：段磊[1] 唐常杰[1] 杨宁[1] 左劼[1] 王悦[2] 郑皎凌[3] 徐开阔[4] 

机构地区：[1]四川大学计算机学院,成都610065 [2]北京大学信息科学技术学院,北京100871 [3]成都信息工程学院软件工程系,成都610225 [4]成都信息工程学院计算机系,成都610225

出　　处：《计算机学报》2011年第10期1831-1842,共12页Chinese Journal of Computers

基　　金：国家自然科学基金(60773169);"十一五"国家科技支撑计划(2006BAI05A01);高等学校博士学科点专项科研基金(20100181120029);四川大学青年教师科研启动基金(2009SCU11030)资助~~

摘　　要：干预规则挖掘是近年从干预实践中提出的新型数据挖掘任务,旨在利用数据挖掘技术探测干预事件,发现最佳干预时机和力度,提供促进事物向期待状态转化的决策支持.文中以四年的研究实践为背景,介绍干预规则挖掘的研究沿革和现状,给出了干预规则挖掘的任务分类.从三个角度,即干预效果预测、干预方法发现和未知干预探测三方面,介绍干预规则挖掘的研究问题、困难和成果.展望了干预规则挖掘未来研究方向.Intervention rule mining is an emerging data mining task,which is derived from the practice of intervention application.It aims at applying data mining techniques on detecting intervention events,discovering the best intervention time and intensity,and decision support for converting objects from undesirable state to desirable state.This paper introduces the research background,as well as the major related advances on intervention rule mining based on the four-year practice,and defines the task classification.Moreover,this paper surveys the research issues,difficulties and achievements in three aspects,i.e.intervention effect prediction,intervention method discovery,and unknown intervention event detection.Finally,this paper discusses the future work of intervention rule mining.

关 键 词：数据挖掘 干预规则 流数据 不确定数据 时间序列 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]