基于Can树的关联规则增量更新算法改进

如何从动态数据中挖掘关联规则是目前知识发现中的一个研究热点。Can树是基于CATS树改进后提出的解决关联规则增量挖掘的一种有效算法，它要求事务中的每个项按照某种特定顺序进行排序后再构建Can树，其顺序一般采用字典序、字母序等。然而，Can树所使用的排序方法有可能使得Can树的规模过大，从而使得算法效率较低。针对该问题，在现有Can树挖掘算法的基础上，使用数据量排序替代现有排序方法，提出了一种基于数据量排序的Can树，并基于新的Can树对原有Can树的建树和挖掘方法进行优化。该方法可以有效减小Can树的规模，实现频繁项集挖掘在空间效率和时间效率上的优化。实验结果表明，该方法在空间效率和时间效率上好于现有的Can树算法，同时具有较好的稳定性。

Improved incremental updating algorithm of association rules based on Can-tree

How to mine association rule from dynamic data is one of the hottest topics in knowledge acquisition. Can-tree, which is based on CATS tree, is an effective algorithm for incremental mining association rule. It requires that all items should be in specific order before building Can-tree, like lexicographic order or alphabetical order. However, the current used ordering method may cause the size of Can-tree too large, thus decreasing the efficiency of the algorithm. Aiming at this problem, this paper proposes a new Can-tree, which sorts items by data size instead of the current ordering methods. Meanwhile, the tree building and mining methods are optimized based on the proposed Can-tree. This method can effectively minimize the size of Can-tree, and improve the efficiency of mining frequent itemset in space and time. The experimental results show that this method outperforms the existing Can-tree algorithm in space efficiency, time efficiency, and stability.