电子学报 ›› 2015, Vol. 43 ›› Issue (2): 269-275.DOI: 10.3969/j.issn.0372-2112.2015.02.010

• 学术论文 • 上一篇    下一篇

广义无冗余情节规则抽取方法研究

尤涛, 徐伟, 杨凯, 杜承烈, 钟冬   

  1. 西北工业大学计算机学院, 陕西西安 710129
  • 收稿日期:2013-12-13 修回日期:2014-08-07 出版日期:2015-02-25
    • 作者简介:
    • 尤 涛 男,1983年生于河南三门峡.西北工业大学计算机学院讲师,研究方向为分布式数据流处理. E-mail:youtao@nwpu.edu.cn;徐 伟 男,1989年生于江苏南京.西北工业大学计算机学院硕士研究生,研究方向为数据流处理.
    • 基金资助:
    • 国家自然科学基金 (No.61303225); 航空科学基金 (No.20135553034); 中央高校基本科研业务费专项资金 (No.3102014JSJ0008)

Research on Extracting Generalized Non-Redundant Episode Rules

YOU Tao, XU Wei, YANG Kai, DU Cheng-lie, ZHONG Dong   

  1. Department of Computer Science and Engineering, Northwestern Polytechnical University, Xi'an, Shaanxi 710129, China
  • Received:2013-12-13 Revised:2014-08-07 Online:2015-02-25 Published:2015-02-25

摘要:

情节规则挖掘旨在发现频繁情节之间的因果关联,现有无损情节规则挖掘方法没有考虑多规则间的关联关系,故而存在大量冗余.利用演绎推导特性对情节规则间的关联关系进行建模,引入无冗余情节迹规则的概念,分析了情节迹冗余的原因,通过最大重叠项冗余性检查给出广义无冗余情节规则抽取算法;证明了广义无冗余情节规则对情节规则的等价表达能力.理论分析和实验评估表明该算法在处理效率基本不变的前提下,提高了情节规则的生成质量.

关键词: 事件序列, 演绎, 情节迹, 最大重叠项, 情节规则

Abstract:

Aiming at the problem that current nondestructive episode rule mining algorithms don't consider the relationship between episode rules and generate redundancy,we model the relationship among the episode rules by using deduction characteristic,and introduce the concept of non-redundant episode trace rules.We also analyze reasons for episode trace redundancy,and present the generalized non-redundant episode rules mining algorithm based on the redundant checking on maximum overlap items.Then we prove that generalized non-redundant episode rules keep the equivalent expression ability to episode rules.Theoretical analysis and experiments demonstrate this algorithm improved the quality of generatedepisode rules with almost the same efficiency.

Key words: event sequence, deduction, episode trace, maximum overlap items, episode rule

中图分类号: