本文基于中药药理文献数据，以中药的药性、药味、毒性、归经、主治、功效6 个属性的所有值作为属性集，使用支持向量机对187 个中药药理作用分别构建预测模型，并使用交叉验证的方法确定模型预测的准确率。然后使用准确率大于90%的中药药理作用预测模型对《中国药典》（2010 版）收录的624味中药进行药理作用的预测。结果发现有108 个模型的准确率大于90%，而抗菌作用预测模型的准确率达到99.76%。对中药的预测中预测百分比最高的是北豆根的抗氧化作用。

Based on Chinese medicine pharmacological literature data, the property, flavor, toxicity, meridian tropism, efficacy, and clinical application of Chinese herbal medicine were used as a set of attributes. The support vector machine(SVM) was used in the establishment of predictive models of 187 pharmacological effects of Chinese herbal medicine respectively. And the cross-validation method was used to determine the accuracy of predictive models. After that, the predictive models with the predictive accuracy rate greater than 90% were used to predicate pharmacological effects of 624 herbals recorded in the Chinese Pharmacopoeia(2010 edition). It was found that the accuracy rate of 108 models was greater than 90%, and the accuracy rate of antibacterial effect predictive model was 99.76%. The highest predictive value of Chinese herbal medicine was the anti-oxidation effect of Menispermi Rhizoma.

科学技术部国家“863”计划项目（2006AA01A122）：中国中医药科学数据网格服务应用，负责人：崔蒙；中国中医科学院基本科研业务费自主选题项目（ZZ050305）：基于支持向量机的中药组分药理作用预测模型的研究，负责人：雷蕾。