中文摘要：

提出一种新的基于DT-KSVM（decision tree kernel support vector machine,决策树支持向量机）的业务感知算法,利用ReliefF特征选择算法提取特征,提出样本间类别可分度计算方法排序不同业务感知难度,优先感知易分业务。在实际网络业务数据集上与传统一对一（one-versus-one）SVM感知方法进行对比,结果表明该方法具有较高的业务识别准确率和更好的时间性能。

英文摘要：

A novel service-aware method based on decision tree kernel support vector machine （DT-KSVM） algorithm was proposed. A service-aware model was developed by using ReliefF algorithm to extract service characteristics, and proposing the separable degree between samples to simply the service-aware process. Through experiment comparison between this proposed model and traditional one-versus-one SVM method, it is shown that the proposed method has a better service-aware accuracy and time performance.