Automatic Detection of Argument Components in Text Using Multinomial Nave Bayes Clasiffier

Abstract

Arguments are often found in various text data, for example in news, essays and articles. Argumentation Mining is a method that automatically identifies argument structures in text documents. This argument structure consists of several components that are very useful for information retrieval and processing information. In this study, a model will be built to auto-matically detect the component of argument, by using naive bayes classifer multinomial, the model will classify argument components into two classes, namely claim and premise. The evaluation uses k-fold cross validation. The most optimal result of this study is the average accuracy of 70.39 % and the average f1-score of 80.42 % with feature extraction, preprocessing and weighting words.