Abstract: In the categorization framework for CVE,this paper designs and constructs a learning model based on SVM,so that it can categorize the CVE according to the different taxonomic features.In the process of constructing a learning model based on SVM,first of all,the training data is generated according to the different taxonomic features in the several vulnerability databases,then a data fusion and cleansing process are designed to eliminate the inconsistencies of data,and finally the n-fold cross-validation method is used to evaluate the effect of the model.The learning model has been verified to have better performance of CVE classification.

Key words: support vector machine (SVM), common vulnerabilities and exposures (CVE), taxonomic feature,classification accuracy