Curious to Click It?-Identifying Clickbait using Deep Learning and Evolutionary Algorithm

Massive outreach of the online media along with changing information consumption patterns have revealed the dark side of digital media. The intentional use of tempting, eye-catching, exaggerated and misleading content to capitalize on the voracious appetite of the readers by creating an information gap has flooded the news websites. Thus, we were motivated to develop deep learning models that utilize the lexical as well as semantic features of the headline and the corresponding text to effectively detect clickbait. The Bidirectional Long Short-Term Memory model using GloVe embedding achieves an accuracy of 98.78% that outperforms the previous work. Furthermore, our study for using the Genetic algorithm for hyperparameter optimization also gave promising results with an accuracy of 95.61%.