【作者】 史东辉； 蔡庆生； 张春阳；

【机构】 中国科学技术大学计算机科学系；

【摘要】 <正>1引言数据挖掘的任务是从大量数据中挖掘出有用的、新奇的、易于理解的知识。现有的很多数据挖掘方法如关联规则,可以有效地从数据中发现大量有用的规则,帮助用户进一步理解数据的真实特性。但是,对于不同的挖掘任务及更多还原

【Abstract】 For some data mining tasks,the existing methodology can seldom acquire the satisfied results. A kind of new multi-strategy approach of data mining is put forward in the paper.Firstly,we convert all kinds of non-categorical data into categorical data,and cluster the categorical data into several clusters.Secondly,generate rules from these subsets,which is different from KDD pre-processing step.At last,compare with rules each other.The focus of multi-strategy approach is to obtain the dissimilarity rules which include important information.In such a manner,we can gain the rules which can not be found in only one strategy.The experiment in this paper demonstrates that the multi-strategy approach is effective.更多还原