Supervised machine learning approach for subjectivity/objectivity classification of social data

Abstract

With the exponential growth of critiques available on social media, the distinction between subjective (i.e., emotional terms) and objective (i.e., factual terms) information is a non-trivial natural language processing task. Therefore, we focus in this paper on the subjectivity/objectivity classification of social data precisely Facebook which is the most popular social network. The proposed approach is based on supervised machine learning techniques. Similar to any supervised application, our approach is composed of three main steps which are features extraction, training and prediction. Our main contribution in this work is to introduce new features such as ontology-based N-grams feature which represent good a indicator for predicting subjective text. Multiple classifiers like Naive Bayes, Support Vector Machine, Random Forest and Multi Layer Perceptron are applied on our data set. The experimental results show that the Random Forest model achieved the highest performance metrics 0.676 of accuracy, 0.697 of precision, 0,677 of recall and 0.683 off-score.