Machine Learning Algorithms for Natural Language Processing Tasks: A Case of COVID-19 Twitter data (Thailand)

Abstract

This paper presents the use of natural language processing for the problem of information extraction and sentiment analysis. The dataset is from Twitter that has the information of people mentioning about COVID- 19, this study has two tasks: (i) classification approach for information extraction task and (ii) deep learning approach for sentiment analysis task. In information extraction task, the data was gathered from twitter that related to COVID-19 information, and the sequence labelling method applied to classify text before giving it to classification algorithms (K-NN, Naïve Bayes, Decision Tree, Random Forest, and SVM). In sentiment analysis task, data was classified by convert the word into index and using word embedding, then to process deep learning algorithm (Bi-directional GRU). The accuracy of two tasks are 98% and 79% respectively.