Clickbait detection in news headlines using RoBERTa-Large language model and deep embeddings.

Area of Science:

• Artificial Intelligence
• Natural Language Processing
• Digital News Analysis

Background:

• Clickbait headline detection is a challenging research area within digital news analysis.
• Existing studies primarily use traditional Machine Learning (ML) and Deep Learning (DL) models.
• There is a need for advanced models to capture complex linguistic nuances.

Purpose of the Study:

• To introduce RoBERTa-Large, a transformer-based architecture, for automated clickbait headline detection.
• To evaluate the effectiveness of RoBERTa-Large against state-of-the-art ML and DL approaches.
• To enhance model interpretability using Explainable AI (XAI) methods.

Main Methods:

• Utilized RoBERTa-Large, a transformer architecture with a self-attention mechanism.
• Employed diverse textual features: TF-IDF, Part-of-Speech tagging, n-grams, word2Vec, FastText, and Sentence Embeddings.
• Evaluated classification performance against established ML and DL models.
• Applied LIME and SHAP for Explainable AI (XAI).

Main Results:

• RoBERTa-Large achieved a classification accuracy of 97%.
• The model significantly outperformed existing ML and DL approaches.
• XAI methods provided insights into the model's decision-making process.

Conclusions:

• RoBERTa-Large demonstrates superior performance in clickbait headline detection.
• Transformer-based models offer advantages in capturing contextual and semantic information.
• Explainable AI enhances the trustworthiness and understanding of automated news analysis systems.