基于改进的卷积神经网络的中文情感分类

Chinese text sentiment classification based on improved convolutional neural networks

探究了基于卷积神经网络的句子级别的中文文本情感分类,模型以文本经过预处理后得到的词向量作为输入。传统的卷积神经网络是由线性卷积层、池化层和全连接层堆叠起来的,提出以跨通道卷积层替代传统线性卷积滤波器,对基本的卷积神经网络进行改进,提高网络的表达能力。实验表明,改进后的卷积神经网络在保证训练速度的情况下,识别率达到91.89%,优于传统的卷积神经网络,有较好的识别能力。

A method of sentiment classification based on convolutional neural networks for Chinese comments, which is expressed by pre-train word vectors, is presented. Classic convolutional neural networks is stacked by convolutional layers,pooling layers and fully connected layer. An improved convolutional neural networks in which a cascade cross channel convolutional layer replaces the traditional linear convolutional filter is proposed to improve and enhance the generalization of the network. The experimental results show that the improved convolutional neural networks achieves better performance with the recognition rate of 91.89% and an acceptable training speed, superior to basic convolutional neural networks.