M82B at Check That! 2021: Multiclass Fake News Detection Using BiLSTM

Abstract

The rapid advancement of web technologies enabled to spread of information online faster. This enabled the spreading of both informative and uninformative information among the people. Fake news represents misinformation which is published in the newspapers, blogs, newsfeed, social media. Fake news is expanding very quickly via social media, generated by humans or machines as well as originating the unrest situation in the society, country. Meanwhile, fake news detection using machine learning is becoming a prominent area in the research to identify the credibility of the news instantaneously. To present this work, we used Bidirectional Long Short-Term Memory (BiLSTM) to predict the news is either fake or true. We participated in the fake news classification shared task of Check That! 2021 workshop. We obtained the dataset from this event to train and evaluate our model. Finally, we were able to achieve 36% model accuracy and 29.0 F1-macro score with our training data.