Bangla Misleading Clickbait Detection Using Ensemble Learning Approach

In this paper, we have concentrated on identifying misleading clickbait headlines in Bangla news articles using an ensemble learning approach. Misleading clickbait headlines are designed to attract readers' attention and encourage them to click on a link, often leading to misleading or sensationalized content. Detecting misleading clickbait is crucial to ensure the credibility and reliability of news sources. The Bangla language, being widely spoken and written, has also seen an increase in misleading clickbait practices, necessitating an effective detection mechanism. Our proposed ensemble learning approach combines multiple transformer models to enhance the accuracy and robustness of misleading clickbait detection. We employ diverse features extracted from news headlines, such as linguistic patterns, sentiment analysis, and semantic embeddings. These features serve as input to the ensemble models, which aggregate their predictions to make a final decision. Our achieved final accuracy is 96 %.