Kernel-based data transformation model for nonlinear classification of symbolic data

Symbolic data are usually composed of some categorical variables used to represent discrete entities in many real-world applications. Mining symbolic is more difficult than numerical due the lack inherent geometric properties this type data. In paper, we use two kinds kernel learning methods create a estimation model and nonlinear classification algorithm for By using smoothing method, construct squared-error consistent probability estimator data, followed by new transformation proposed embed into Euclidean space. Based on model, inner product distance measure between objects reformulated, allowing Support Vector Machine (SVM), called SVM-S, be defined Mercer method. The experiment results show that SVM can much effective based our measures.