Theme-enhanced text sentiment classification method based on VAE and Attention

The invention relates to a topic enhancement text sentiment classification method based on VAE and Attention, and belongs to the field of natural language processing, and the method comprises the following steps: S1, converting preprocessed text data into word vectors; s2, reconstructing BoW input by using a variational auto-encoder (VAE) to learn potential topics and keywords; s3, sentence feature information is extracted through the CNN-Bi-LSTM; and S4, using an attention mechanism Attention to calculate a vector output by the last time sequence of the CNN-Bi-LSTM layer and a weight of a potential topic vector, carrying out weighted summation to obtain a feature vector, and then carrying out softmax classification. 本发明涉及一种基于VAE和Attention的主题增强文本情感分类方法，属于自然语言处理领域，包括以下步骤：S1：将预处理后的文本数据转化为词向量；S2：使用变分自编码器VAE重建BoW输入来学习潜在主题和关键字；S3：通过CNN-Bi-LSTM提取句子特征信息；S4：利用注意力机制Attention计算CNN-Bi-LSTM层最后一个时序输出的向量与潜在主题向量的权重进行加权求和作为特征向量，再进行softmax分类。