Sentiment Analysis Of Weibo Based On Machine Learning

With the rapid development of Internet technology,many users use social networks to publish their emotional attitudes in the form of comments or feedback.Analyzing users' emotional state from data has become a research hot,which has a great effect on individuals,enterprises and society.Sentiment analysis is the practice of apply natural language processing to identify subjective information from text.This paper studies sentiment analysis at different levels based on emotion dictionary,machine learning and deep learning.The characteristics of sentiment analysis is summarized.We make a thorough study on the construction of sentiment analysis model based on machine learning.Aiming at the potential suicide tendency problem in social media,according to the domain knowledge hierarchical classification scheme,a microblog user suicide tendency prediction model based on hierarchical support vector machine is proposed.By experiments,it shows that the prediction accuracy for suicide probability ups to 0.848,which can effectively predict the suicidal tendency of Weibo users.This model provides early identification of high suicide risk groups and can be used for suicide tendency detection and intervention to reduce suicide probability.At the same time,it is found that there is a relationship between the Weibo publishing time and the suicide risk probability.In order to enhance the applicability of Weibo sentiment analysis in the field,we research the refinement of sentiment classification granularity.Based on Convolutional Neural Network,we propose a fine-grained sentiment classification model(FSCM＿CNN)to effectively learn the emotional features of the texts.On the data set,disclosed from NLPCC evaluation,we carried out experiments at sentence level and document level respectively,this model achieves better performance than the best evaluation results of the conference.