摘要： 针对树挖掘算法产生大量频繁子树和树数据库随时间变化的问题，提出最小频繁闭树增量式更新算法以及增量式更新策略，能充分利用已有挖掘知识，无须重新运行树挖掘算法，并且只需进行一次数据库扫描操作。给出一种候选子树剪枝方法，能减少树同构判别次数，有效提高算法的运行效率。通过大量实验结果表明，该算法有效可行且效率较高。

关键词: 数据挖掘, 有序树, 频繁子树, 频繁闭树, 增量更新

Abstract: Tree mining algorithm always produces a lot of problems such as frequent subtrees and tree database changing over time. This paper proposes least frequent closed tree mining algorithm and incremental updating algorithm for frequent subtrees. It proposes incremental strategy, makes full use of exists data, without re-running tree mining algorithm during update mining, and needs scaning database only once. It proposes tree pruning method, which can lessen the time of distinguishing isomorphism, and improve the efficiency of algorithm working. The final adoption of a large number of experiments shows that incremental updating algorithm proposed in this paper is effective and feasible.

Key words: data mining, ordered tree, frequent subtree, frequent closed tree, incremental updating

中图分类号: 

• TP311