Computational Argumentation for xAI

This track focuses on the emerging field of Computational Argumentation and its intersection with Explainable AI, exploring novel approaches that harness the power of argumentation to enhance the interpretability and explainability of AI systems. Examples include implementing interpretable machine learning models using rule-based approaches and expressing rules in a structured argumentative form, as well as models that incorporate feature importance explanation through argumentation by providing arguments for and against the relevance of each feature. Other examples include using argumentative structures for visual explanations, such as interactive graphs or diagrams, representing a decision-making process. Argumentation can also help identify biases within a model by evaluating the supporting and opposing evidence for different predictions or improving performance via error analysis. When a model makes errors, argumentation can be used to analyze and present the reasons behind those errors. Argumentation can be employed to create interactive explanation interfaces, allowing users to interact with a model’s explanations and challenge or question its decisions. Eventually, argumentation can be adapted to incorporate expert opinions and domain-specific knowledge into the AI model. This track invites submissions contributing to this interdisciplinary domain’s theoretical foundations, methodological advancements, and practical applications. Researchers and practitioners are encouraged to share their insights, methodologies, and findings, fostering a collaborative environment to propel the field forward as a tangible tool for Explainable Artificial Intelligence.