Research On Sentiment Analysis Methods Based On Big Data

The development of the Internet and the continuous increase of smart terminals have brought new vitality to the development of all walks of life in society.Driven by the Internet and mobile Internet,Weibo has developed rapidly.With the continuous increase of Weibo users,Weibo has become one of the largest online social networking platforms in China.Through the mining of Chinese Weibo content,the commercial value of Weibo can be discovered.Weibo users exchange emotions and tell their emotions on Weibo have become more and more normal.In recent years,with the continuous improvement of the material quality of life,people pay more and more attention to the sublimation of spirits.Depression as a mental illness has also received increasing attention from the society.It is helpful to determine whether the Weibo users have depression,and it is also helpful for the discovery of depression through the analysis of depression emotions in Weibo.This paper briefly introduces the definition of sentiment analysis and the research status of domestic and foreign sentiment analysis in the field of Weibo.It also introduces the related technologies of sentiment analysis,including dictionary-based technology and machine learning-based technology.Then carries out related research on these technologies.Through the analysis and research of Sina Weibo,based on the existing several famous lexicons,a dictionary of depression emotions is constructed,and degree adverbs dictionary and negative word dictionary are adjusted as needed.And through the analysis of the characteristics of Weibos issued by Weibo users suffering from depression,the dictionary is expanded to obtain the final dictionary of depression emotions.This paper makes use of two kinds of classification models,Support Vector Machine(SVM)and Naive Bayes(NB),in machine learning.Experiments show that the use of Support Vector Machine classification model to classify text is more effective than simple The Bayesian classifier has a better effect.Both methods need to annotate the corpus.Therefore,using the Depressive Emotion dictionary to analyze the Weibo users' Weibo content has certain advantages.This paper also uses the dictionary of depression emotions to find the emotional scores of each Weibo sent by Weibo users based on the relationship between words,and to record the number of Weibos without depression and the number of Weibos containing depression,and then calculate them.The ratio is finally compared with the given threshold to determine the possibility of Weibo users having depression.