一种基于序列挖掘的分类系统框架  

Classification System Framework Based on Sequence Mining

作　　者：原野[1] 沈钧毅[1] 

机构地区：[1]西安交通大学电子与信息工程学院,西安710049

出　　处：《西安交通大学学报》2004年第4期400-403,共4页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(60173058).

摘　　要：为了有效地对序列数据进行分类,提出了一种集成分类挖掘和序列模式挖掘技术的分类系统框架(SPACS).先采用一套约束和裁减策略,为每个分类挖掘频繁序列模式,并将其转换为分类序列规则(CSR);再利用平均CSR匹配置信度和一个规则匹配算法构建有效的序列数据分类器.SPACS不需要在提取序列的特征后采用传统方法进行分类,可以直接利用从序列数据中提取出的频繁序列进行分类.实验结果表明,对于序列类型的数据的分类,SPACS比传统的决策树和关联分类方法具有更高的分类精度.In order to classify the sequential data effectively, a classification system framework based on the integration of classification and sequential pattern mining (SPACS) is proposed. A set of constraining and pruning strategies are used to mine all the frequent sequential patterns for each class. These sequential patterns are transformed into class sequence rules (CSRs). An effective sequential data classifier is constructed using average CSR matching confidence and a rule-matching algorithm. SPACS does not need use the traditional methods to classify data after extracting sequence features. It can classify the sequential data directly by using the frequent sequences. The experimental results show that SPACS has higher classification precision than the traditional decision tree and associative classification methods.

关 键 词：序列模式挖掘 分类 分类序列规则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]