将范例推理中的范例初步匹配看作文本分类的特殊情形,提出基于类别中心向量的分类算法。通过确定待处理案例的归属类别,缩小范例检索范围,减少在范例精确匹配阶段的计算量,提高案例初步匹配的准确性。在此基础上,将上述算法应用在对交通事故案例的处理与交通信息预警系统中。实验与使用表明,该算法能较为准确地判断事故类型并给出相应的预警信息。 Taking the example preliminary matching in example reasoning as a special case of text categorization, a classification algorithm based on category center vectors is proposed. By determining the classification of the cases to be processed, narrowing the scope of sample retrieval, reducing the amount of computation in the sample exact matching phase and improving the accuracy of initial case matching. On this basis, the above algorithm is applied to the handling of traffic accidents and traffic information early warning system. Experiments and applications show that the algorithm can accurately determine the type of accident and give the corresponding warning information.