Entiment Analysis Of Comment Text Based On Deep Learning

This article starts with the topic of "Emotion Classification of Comment Texts Based on Deep Learning".First,it conducts an in-depth investigation on the current research status and commonly used algorithms at home and abroad,and briefly introduces and analyzes the current mainstream sentiment analysis methods.As a direction of machine learning,deep learning has become a hot research topic in emotion classification in the field of natural language processing.This paper uses deep learning models to study the sentiment classification problem of Chinese text and cross-domain sentiment classification problem.The main research contents are as follows:This paper first proposes a hybrid serial network structure of Bi-LSTM network and CNN network.First,the Bi-LSTM model is used to extract the text information of the context of the text,and an attention mechanism is added to the model to solve the problem of different contributions of different words in the text.At the same time,for cross-domain tasks,the source domain dataset is used to train the BiLSTM-ACNN model and save the model weight information,and then use a small amount of data set in the target domain to fine-tune the model,update part of the weight information,and realize the cross-domain migration of the model.The final experimental data shows that the BiLSTM-ACNN model can effectively improve the performance of cross-domain sentiment classification of Chinese datasets when processing sentiment classification tasks and cross-domain sentiment classification tasks.In the traditional classic neural network model,the feature dimension is too high,and the loss of feature information in the pooling layer leads to the loss of details of emotional vocabulary.This paper draws on the unique advantages of capsule network in feature extraction,and uses the capsule network model based on convolutional neural network to study the task of text sentiment classification.The model uses multiple convolution kernels to extract text features to generate multiple feature maps to enrich the first-level representation.Then connect the capsule network with dynamic routing mechanism.The neuron activity in the capsule network capsule can represent the attribute information of the words in the text,and it can also extract the structural information of the text.Experimental results show that the deep learning model can extract features from the original data,and the dynamic routing mechanism of the capsule network can better obtain the text feature capabilities,and it has a better performance in the emotion classification task.