【摘要】 目前大多数频繁子树算法都是挖掘频繁子树完全集,这些算法数据搜索空间的内存开销和输出的结果集都非常庞大。为了减小结果集,提出基于子树约束的最大频繁子树算法——CSMTreeMiner,采用垂直和层次扩展的方法来枚举频繁子树,并使用覆盖关系来对不可能生成最大频繁子树的模式进行删除。实验结果验证CSMTreeMiner算法的有效性和稳定性。更多还原

【Abstract】 Most algorithms for mining frequent subtrees are mining completed frequent sets, so the searching space and results are very huge. For reducing the results’ number, proposes CSMTreeMiner, a maximal frequent subtree mining algorithm based on subtree constraint. This algorithm uses vertical and level extension technique to enumerate subtrees, and uses the cover relationship to delete the pattern which is impossilbe to generate maximal frequent subtree. Experimental result shows that CSMTreeMiner is effective and stable.更多还原