Research On Chinese Sentiment Classification Based On CNN

Sentiment classification is a research hotspot in the field of natural language processing in recent years.It is also an important branch of text data mining and has received more and more scholars' attention and research.With the development of various social media and e-commerce,people began to express their opinions on social networks,and these data have become important Internet resources.There are comments about hot events in the massive texts published,reviews of goods or services,reviews of movies or books,and most of these comments have obvious sentiment tendencies.How to excavate the subjective information with emotions and likes and dislikes from a vast amount of text is the work of emotion classification.Sentiment classification is widely applied in product analysis and recommendations.In recent years,the rise of deep learning has made a breakthrough in the field of natural language processing,and the convolution neural network has been widely used in text classification.With the popularity of e-commerce,more and more appeared on the network information comments goods,more than the traditional methods for dictionary based on emotion,or machine learning,etc.,these problems data sparse method,and can't dig the deep semantic relations between the word and the word,this thesis puts forward the convolutional neural network model to solve the problem of Chinese comments emotion classification,and introduces Word2 vec term vectors,term vectors of each comment can be converted to emotional input matrix in convolution neural network are classified,improve emotion classification accuracy.In order to compare the performance of the proposed convolutional neural network,the long and short term memory neural network and the support vector machine model were used as the baseline methods.Through contrast experiment,convolution neural network classification performance is better than the baseline method,the experimental results show that: the convolution neural network to extract features of different dimensions and stable performance can better dig out the hidden emotional information between term vectors;Although long-term and short-term memory neural network can preserve long-term memory information,it will cause the loss of important information in the training process,so its performance is not as good as convolution neural network;the SVM model ignores the word order and semantic information between words,which leads to poor performance.