Semantic Analyses using NLP for Twitter Posts in Different Languages — An Example of Sentiments on Global Warming Issues
- Natural language processing (NLP) can be adopted to teach machines to understand human languages. However, when carrying out semantic analyses using NLP, differences exist in the preprocessing steps and labeling methods for social network users’ posts written in different languages. Instead of carrying out multiple NLP tasks for each language, this study intends to explore some consistencies in data preprocessing and labeling methods to analyze posts written in multiple languages and identify the users’ emotions during posting. After a sufficient number of Twitter posts written in Chinese, English, and Japanese were collected, NLP techniques were conducted followed by establishing a convolutional neural network to verify the accuracy of evaluating Twitter users’ emotional status (worried, suspicious, warning) during posting. The same steps of conducting semantic analyses using NLP were found from this study’s initial findings for multi-language posts such that the posts were effectively labeled along with accurate categorization of users’ emotions.
Keywords - Natural Language Processing, Semantic Analysis, Labeling, Global Warming, Convolutional Neural Network