Research On Sentiment Analysis Based On Convolution Neural Network And Minimal Gated Unit

In recent years,e-commerce platforms have sprung up like mushrooms after a spring rain,and people’s dependence on online shopping platforms has become higher and higher.At the same time,the comment texts on the platforms have also shown an explosive growth.These texts contain rich emotional content,and for merchants and other consumers,collecting and analyzing this data can bring them high reference value.As an important part of the field of natural language processing(NLP),text sentiment analysis has been widely concerned by scholars at home and abroad.In sentiment classification tasks,the features extracted by traditional convolutional neural networks(CNN)have limitations,and often ignore the context information of words,while the more commonly used long short-term memory network(LSTM)also has the problem of too many parameters.Aiming at the shortcomings and deficiencies of deep neural networks in sentiment classification tasks,this paper proposes a CNN-BiMGU model and a PAPC-BiMGU model,respectively.The CNN-BiMGU model is designed with two parallel channels,CNN channel and MGU channel.The CNN channel contains three sub-channels.The size of the convolution kernel in each channel is different,which can dig out the multi-level feature content in the text.The MGU channel uses a combination of forward MGU and backward MGU,and mainly extracts the feature content implicit in the text.The PAPC-BiMGU model introduces an attention mechanism on the basis of the CNN-BiMGU model.First,the input text is divided into the text content based on the evaluation attributes,and then the general Attention mechanism is improved from the perspective of word part of speech and position.Important features in the text are given more attention,resulting in more accurate classification results.In this paper,the CNN-BiMGU model and the PAPC-BiMGU model are compared with other classification models.Experiments show that the models proposed in the paper have achieved good results in processing sentiment classification tasks,and the PAPC-BiMGU model is more prominent in multi-attribute texts.Finally,according to the proposed model,an emotion classification test system is constructed...