基于特征族群语义扩散核的半监督农业文本分类

Semi-supervised Agricultural Text Classification Based on Feature Cluster Semantic Diffusion Kernel

农业文本分类旨在对主流的农业信息网抽取的文本数据集进行分类.在样本充足的情形下,经典的支持向量机方法能取得较好的效果,然而在样本较少或者样本矩阵很稀疏的情形下效果较差.提出了一种基于特征族群语义扩散核(它是语义扩散核的一种)和支持向量机的半监督农业文本分类方法.该方法在经典的支持向量机方法基础上结合特征族群语义扩散核,使得农业文本分类准确率得到一个显著的提升,在训练集样本数量只有原来一半的数量情况下预测原来的测试样本,预测准确率几乎与原来的相同.

Agricultural text classification aims to classify the text datasets extracted from mainstream agricultural information networks. With sufficient samples ,the classical support vector machine （SVM） method can achieve good results ,but the classification performance is usually unsatisfactory when the number of the samples is small or the sample matrix is sparse. This paper presents a semi - supervised agricul- tural text classification method which is based on the feature - cluster semantic diffusion kernel and SVM. Combined the classical SVM with feature -cluster semantic diffusion kernel, this method yields a significantly better classification results than those obtained by the classical method. More specifically,when the number of the training samples is reduced to the half of that of the original training set,the prediction accuracy of the proposed method is almost the same as that of the classical method with the original training set.