What Make Facts Stand Out from Opinions? Distinguishing Facts from Opinions in News Media
December 2015
Authors
Abstract
In order to make arguments convincing, often opinions are presented in disguise of facts in news media. It is hence necessary to distinguish facts from opinions in order to get a true picture of the happenings from the news published in media. However, this task in itself is quite challenging as it involves not just associating entities with events and dates but also learning patterns and rules describing facts. Till date, considerable amount of work has been conducted on subjectivity analysis under sentiment analysis but the focus is still much on determining opinions rather than facts. This paper, which has a larger focus on determining facts, first deals with the definition of facts and then moves on to the discussion the linguistic patterns of some representative examples of facts from the corpus of editorials. Then, we present and discuss two frameworks for determining facts. Finally, we present the performance accuracies of the three Machine Learning methods for the given task, namely Multinomial Naïve Bayes, Logistic Regression and Support Vector Machine (SVM).