A Short Text Sentiment Analysis Model Integrating Output Features of Bi-LSTM and CNN
For short text sentiment analysis tasks,short texts have short language expressions,sparse semantics,and carry few features,and general network models learn insufficient semantic features from short texts.In response to the above problems,this paper proposes a short text sentiment analysis model that fuses the output features of Bi-LSTM and CNN.The text representa-tion is extracted by Bi-LSTM and CNN respectively,and then the contextual semantic features obtained by the Bi-LSTM layer and the deep abstract semantic features obtained by the CNN layer are fused as the final classification features.Finally,the effectiveness of the model is verified on the review corpus in Tan Songbo Hotel and the SST-2 corpus.
万方数据
维普
