Counterfactual Explanations for Commonly used Text Classifiers focus on Review Classification

In recent years, the use of AI-driven decisionmaking systems-based applications has increased significantly, and many of these applications involve making critical decisions that have a significant impact on human lives. Even though AI made systems are automated and easier to use, but lack of transparency and interpretability has become a major drawback of it. That causes most of the machine learning models that are used for decision-making processes are black box, which means the internal behaviors of the models are more complex and difficult for humans to understand. With the implementation of the European Union's GDPR, the area of Model explainability is being raised as a hot topic and it is the next step of AI. Explainable Artificial intelligence provides an explanation for the complex machine learning processes into insights that are easily comprehensible and provide transparency into the decisionmaking processes of AI systems. This research is about the development of a novel counterfactual rule generation-based explainable AI solution for the text classification domain, focusing on K nearest neighbor, logistic regression, random forest, and support vector machine models. The novel explanations of the KNN and SVM models are based on a customized word-flipping generator. Further explanation of Logistic regression provides a prediction score-based method, and Random Forest provides a feature importance-based method.