| 数据流挖掘技术 |
| |
| 引用本文： | 姜远,刘力平. 数据流挖掘技术[J]. 江南大学学报(自然科学版), 2007, 6(6): 654-657 |
| |
| 作者姓名： | 姜远 刘力平 |
| |
| 作者单位： | 南京大学,软件新技术国家重点实验室,江苏,南京,210093 |
| |
| 基金项目： | 国家自然科学基金项目(60505013)，江苏省自然科学基金项目(BK2005412) |
| |
| 摘 要： | 简要概述了数据流挖掘技术,探讨了数据流的特点.数据流的概念漂移现象,给数据流上的数据挖掘带来很大困难.由于计算机的内存有限,数据窗口技术只针对最近的数据,而最近的数据常常导致数据挖掘系统中的分类器过配,文中介绍了解决这一问题的方法,并讨论了数据流挖掘技术的应用.
|
| 关 键 词： | 数据挖掘 数据流 滑动窗口 分类 |
| 文章编号： | 1671-7147(2007)06-0654-04 |
| 修稿时间： | 2007-05-05 |
Survey of Data Mining on Data Stream |
| |
| JIANG Yuan,LIU Li-ping. Survey of Data Mining on Data Stream[J]. Journal of Southern Yangtze University:Natural Science Edition, 2007, 6(6): 654-657 |
| |
| Authors: | JIANG Yuan LIU Li-ping |
| |
| Abstract: | The paper briefly introduces the techniques of data mining on data stream.The characteristics of data stream,such as unbounded,fast-flowing,continuous,and bursty are discussed.The concept in the data stream often changes as time passes.Sometimes,concept conflicts exist in the data stream,which brings a lot of difficulties to mining on data stream.Nowadays,limited memory can not satisfy the requirements of constantly increasing data."Sliding window" technique which only considers recent data in data stream is such a kind of technique to meet the new requirements.But in such technique overfitting often occurs.Some algorithms are proposed to cope with such problems.This paper gives a brief introduction to these algorithms.Some applications of data mining on data stream are also discussed. |
| |
| Keywords: | data mining data stream sliding window classify |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
