Explainable Artificial Intelligence for Sarcasm Detection in Dialogues

Sarcasm detection in dialogues has been gaining popularity among natural language processing (NLP) researchers with the increased use of conversational threads on social media. Capturing knowledge domain discourse, context propagation during course dialogue, and situational tone speaker are some important features to train machine learning models for detecting sarcasm real time. As comedies vibrantly represent human mannerism behaviour everyday real-life situations, this research demonstrates an ensemble supervised algorithm detect benchmark dialogue dataset, MUStARD. The punch-line utterance its associated taken as eXtreme Gradient Boosting (XGBoost) method. primary goal is predict each using chronological nature a scene. Further, it vital prevent model bias help decision makers understand how right way. Therefore, twin research, we make used interpretable. This done two post hoc interpretability approaches, Local Interpretable Model-agnostic Explanations (LIME) Shapley Additive exPlanations (SHAP), generate explanations output trained classifier. classification results clearly depict importance capturing intersentence threads. methods show words (features) that influence most user making dialogues.