1.文本预处理:对原始数据集进行清洗、分词、去停用词;
2.特征构建:对文本预处理后构建词向量库(word2index和index2word);
3.模型选择:先后运用TextCNN、LSTM、TextRCNN四种深度学习模型训练
CNN: loss: 1.0396 - accuracy: 0.6423 - val_loss: 1.1583 - val_accuracy: 0.5306
LSTM: loss: 0.9357 - accuracy: 0.6418 - val_loss: 1.3736 - val_accuracy: 0.4895
RCNN: loss: 1.1153 - accuracy: 0.5993 - val_loss: 1.4426 - val_accuracy: 0.4012
TensorFlow版本:TensorFlow2.0