Abstract: This paper introduces a feasible method formalizing the calculation of information of a prescription of Chinese medicine，and discusses quantified representation of attribute characteristics in prescription samples.Based on aforesaid method and representation，prescription samples are classified by pattern recognition system on the basis of SVM.The acquired results are satisfactory，which are also consistent with or close to the general cognition of principal theory of Chinese medicine.The experience has shown that，machine learning method is a feasible solution to mine thinking mode of Chinese medicine experts when they writing prescriptions.It has bright future of application.

Key words: prescriptions of Chinese medicine, characteristics acquisition, Support Vector Machine（SVM）

摘要： 介绍了一种实现方剂知识信息形式化计算的方法，并进行了量化的表达，在此基础上运用基于SVM的模式识别系统，以“方剂分类”为目标进行实验，其结果符合或接近中医学的普遍认识，获得了满意的效果。实验表明，机器学习技术是挖掘中医处方思维模式特征的可行方法，有很好的应用前景。

关键词: 中医方剂, 特征提取, 支持向量机（SVM）