Emotional Tendency Analysis of Twitter Texts Based on CNNs Model

With the development of mobile communication technology, increasing online social media provides opportunities for people to express their views, which plays an important role in public opinion analysis and monitoring. The main topic of this study is the emotional analysis of Twitter English users. Reptile technology is applied to obtain and clean tweets related to China, after tagging data referred the Dictionary of strong words which is composed of statistical word frequencies, GloVe is used to generate word vectors and convolutional neural networks (CNN) is used to automatically classify tweets. Thus an accurate CNN model is established to analyze the popular tweets related to China from 2014 to 2018. The results show that China occupies an important position in world politics in 2018 with more negative Twitter comments. The number of supporters is the largest in 2014, but it has a sudden decline in 2015 and remains around 45% from 2016 to 2018.