适用于协议特征提取的关联规则改进算法

Improved Association Rules Algorithm for Protocol Signatures Extracting

  • 摘要: 借鉴关联规则挖掘的思想,引入序列项目集的概念,使算法能够处理集合事物和具有序列特性的项目;通过递推的方法依次得出不同长度的特征字段,并利用偏移属性集加以约束去除无效字段,有效控制约束频繁集的规模;最后依据选择策略从约束频繁集中选出最终的特征字段。实验结果表明只要选取合适的参数,用该方法提取协议特征是行之有效的。

     

    Abstract: The notion of sequence itemset is introduced for algorithm to deal with permutations items or transactions of itemsets. A recursive method is provided to generate signatures with different length in turn. The algorithm utilizes offset attributor set to restrict and remove ineffective itemsets. According to selection principle, the optimal constrained frequency sequence itemset as signatures can be found. The result shows that the signatures extracted by this algorithm are reasonable and effective.

     

/

返回文章
返回