Research On Sentiment Classification Of Weibo Text Data Based On Deep Learning Algorithm

Sentiment analysis technology is one of the most popular directions in the field of natural language processing.Its function is to use some rules to extract the text context information,and judges the text author's emotional tendency according to the preset algorithm.Weibo social platform is one of the most popular Chinese social platforms,if the Weibo comment text can be judged in time and accurately,the commercial value is huge.But it is impossible to discern such a large amount of text by hand.The continuous development of sentiment analysis technology makes it possible for machines to help organizations analyze user emotions.From the earliest emotional dictionary approach to the use of machine learning algorithms to identify sentimental tendencies,then the emergence of deep learning technology pushes sentiment analysis technology to a new climax.Deep learning is a collective term for a series of neural network algorithms,including convolutional neural networks,recurrent neural networks,etc.Deep learning is different from previous sentiment analysis techniques;it can learn text information autonomously without relying on manual selection.So deep learning gets a very high achievement in the emotional classification task.However,at present,domestic research on the classification of Weibo emotions is scarce.Most of the past studies use emotional dictionary methods and machine learning algorithms,and the use of deep learning algorithms to study Weibo texts is less,and the algorithm used is a little simple.Through in-depth research on the two representative structures of deep learning algorithms,it is found that the two algorithms have advantages and disadvantages.If the advantages of the two can be extracted to work together,then it is very likely that a better classification result will be produced in the face of classification tasks.CNN algorithm and CNN-LSTM algorithm are based on this idea.Based on this,this paper takes Weibo user "Headline News" as the research object,crawls its 10039 user comments,after data cleaning,uses Python's Jieba library and Word2 vec model for text tokenization and vectorization respectively,and finally constructs a model can be used to judge the sentiment tendency of Weibo review text.This model uses five deep learning algorithms including Convolutional Neural Network(CNN),Long Short-Term Memory Network(LSTM),Bidirectional LongShort-Term Memory Network(Bi LSTM),and fusion algorithms LSTM-CNN and CNN-LSTM.Experiments show that the newly proposed LSTM-CNN algorithm has the best effect on sentiment classification.This shows that the idea of constructing fusion algorithm in this paper is feasible.The research in this article expands the idea of using deep learning algorithms to study Weibo texts,and the new algorithm also provides reference for the development of deep learning algorithms.