计算机应用 ›› 2010, Vol. 30 ›› Issue (05): 1300-1303.

• 数据挖掘与人工智能 • 上一篇    下一篇

一种新的频繁子树增量式更新方法

郭鑫1,黄云2,颜一鸣2,周清平2   

  1. 1. 吉首大学
    2.
  • 收稿日期:2009-11-19 修回日期:2010-01-11 发布日期:2010-05-04 出版日期:2010-05-01
  • 通讯作者: 郭鑫
  • 基金资助:
    湖南省教育厅资助项目

Novel incremental updating algorithm for frequent subtrees

  • Received:2009-11-19 Revised:2010-01-11 Online:2010-05-04 Published:2010-05-01
  • Contact: guo xin

摘要: 讨论频繁子树增量式更新问题,提出一种新的频繁子树增量式更新算法。提出有效树集概念和增量式更新策略,在更新挖掘时,无须重新运行子树挖掘程序,能充分利用已有的挖掘结果,算法只需要进行一次数据库遍历操作。提出候选子树剪枝策略,在更新挖掘过程中,能大幅减少子树同构次数,有效地提高了算法的运行效率。通过大量实验分析表明,算法有效可行且具有较高的运行效率。

关键词: 数据挖掘, 有序树, 频繁子树, 子树同构, 增量更新

Abstract: The incremental update for frequent subtrees was discussed and a novel incremental updating algorithm for frequent subtrees was proposed. The concept of effective tree collection and incremental strategy were put forward, which did not need re-run tree mining algorithm during update mining and could make full use of the existing data, and need scan database only once. Subtree pruning strategy was put forward to reduce the number of subtrees distinguishing isomorphism during update mining, which improved the operational efficiency of the algorithm. The experimental results show that the proposed algorithm is effective and feasible and has significant operation efficiency.

Key words: data mining, ordered tree, frequent subtree, subtree isomorphism, incremental update