Stance Classification Towards Political Figures on Blog Writing

In this paper, we present new application of stance detection task on politic domain. Our goal is to determine whether the writer of the blog article is on the position supporting a political figure to compete and win in a general election event, for example a candidate of President in the Presidential election. We performed the experiment using five different case studies. We examined three baseline machine learning models using combination of n-gram, sentiment lexicon, orthography, and word embedding features. The highest macro-average F1 score was achieved by model trained on Support Vector Machine classifier using a combination of word2vec and unigram features, which is 63,54%.