| Abstract: | Technology advancements, such as social media, are now essential tools for connecting with the rest of the world, including political figures, governments, and social media activists. Recently, people have used social media to express their opinions about a particular subject or target. There are numerous uses for stance classification or detection in the world of NLP. Such as automatic stance recognition of whether a community is for or against a specific point of view in relation to religious and political issues, either in favor or against the stated targets.in this study, we constructed our own dataset with a total of 3,126 comments, of which are targeted Prosperity Party. Once the data has been collected and annotated using annotation guidelines, after that, the data were preprocessed, and morphologically analyzed. Then, we have used 4 different types of feature extraction techniques: BOW, N-gram, TF-IDF and word2vec and we trained three different machine learning algorithms SVM, LR and RF using each feature extraction techniques. according to the results from the experiments, we achieved accuracy score of 0.82 using TF-IDF feature extraction and SVM. based on these results, we draw the conclusion that the political stance classifier performed better classification utilizing feature extraction techniques using TF-IDF and SVM machine learning algorithm. |
