An Effective Approach for Clickbait Detection Based on Supervised Machine Learning Technique
DOI:
https://doi.org/10.3991/ijoe.v15i03.9843Keywords:
Clickbait, F1-score, ROC curve, Similarity, Feature extraction, Formality, Reada-bility, Support Vector Machine (SVM), Data Modeling.Abstract
Clickbait is a term used to describe a deceiving web content that uses ambiguity to prompt the user into clicking a link. It aims to increase the number of online readers in order to generate more advertising revenue. In other words, Clickbait is used to describe a type of hyperlink on a web page which seduces a user to click a link to continue reading a specific article.
Typically such links will forward the visitor to a page that requires payment, registration, or lead a user to a site, which tries to sell user something or possibly extort user, by withholding the promised "bait". We use supervised machine learning to create a model that is trained on 24 features. This method achieved an F1-score of 79% and an area under the ROC curve of 0.7. Our methodology emphasises the importance of using features extracted from different elements of social media posts along with features that are extracted from the title and the article. In this research, we show that it is possible to identify Clickbaits using all parts of the post while keeping the number of features as minimum as possible.
