The workshop aims to bring together researchers, practitioners, and domain experts to exchange knowledge, address challenges, and outline future directions for developing explainable, interpretable, and transparent AI systems. It focuses on advancing Explainable Artificial Intelligence (XAI) by incorporating knowledge and semantics as core components. Contributions will address “demystifying the black-box” nature of AI and tailoring explanations to diverse user expertise levels, supporting equitable and fair decision-making for long-term sustainability. The workshop seeks to overcome the challenges of embedding semantic abstractions into intelligent information systems. These efforts are intended to foster trust, improve debugging, and encourage the adoption of AI in critical domains such as healthcare, finance, policy-making, and education.

Topics of interest include, but are not limited to:

- Foundational Methods for Explainability:
• Model-agnostic and model-specific approaches to explanation.
• Use of explainable and/or interpretable surrogate models for black-box AI systems.
• Techniques for improving transparency in deep neural networks.
- Knowledge-Driven Explainability:
• Integration of knowledge graphs and ontologies for semantic explanations.
• Tools and frameworks for explicit representation of knowledge in AI systems.
• Semantic-level abstractions for user-aligned explanations.
- User-Centric, Human-AI Interaction, and Domain-Specific Explainability:
• Designing user-aligned explanations for diverse expertise levels.
• AI explainability for critical applications such as:
• Diagnostic systems (e.g., Healthcare).
• Autonomous systems (e.g., decision transparency).
• Social systems (e.g., policy-making, ecological transition, or funds allocation).
• Tailored explanations for non-technical users, such as doctors, engineers, and policy-makers.
- Challenges in Black-Box AI:
• Addressing the inherent opacity of deep learning and other black-box models.
• Trade-offs between accuracy and explainability in AI systems.
• Explainability challenges in large-scale, real-world AI deployments.
- Historical and Modern Approaches:
• Lessons from traditional knowledge-based systems on reasoning explanations.
• Advances in description logic, ontologies, and their role in explainability.
• Bridging classical and modern AI paradigms for comprehensive explanations.
- Emerging Techniques and Innovations:
• Explainability in federated and distributed AI systems.
• Methods to enhance transparency in multi-modal AI systems.
• Explainability in generative AI models (e.g., LLMs and diffusion models).
- Explainability Frameworks and Standards:
• Guidelines and best practices for developing explainable AI.
• Standardization of explainability metrics and benchmarks.
• Explainability in the context of ethics, governance, and policy-making.
- Future Directions in Explainability Research:
• Novel paradigms for embedding knowledge into explainable AI.
• Explainability in hybrid AI systems combining symbolic and sub-symbolic methods.
• The role of explainability in advancing human-AI collaboration.
• Cross-disciplinary approaches combining AI with cognitive science and XAI-assisted decision making.


PUBLICATIONS
------------
After thorough reviewing by the workshop program committee, all accepted papers will be published in a special section of the conference proceedings book - under an ISBN reference and on digital support.
All papers presented at the conference venue will be available at the SCITEPRESS Digital Library (http://www.scitepress.org/DigitalLibrary/).
SCITEPRESS is a member of CrossRef (http://www.crossref.org/) and every paper is given a DOI (Digital Object Identifier).

Indexation: The proceedings will be submitted to Thomson Reuters Conference Proceedings Citation Index (ISI), INSPEC, Google Scholar, DBLP, Semantic Scholar, EI (Elsevier Index) and Scopus for indexation.

A short list of papers that underwent additional reviews by independent members of the Conference Program Committee will be considered for post-publication opportunities or best paper awards, with the possibility of having extended versions published in special issues of the Springer Nature Computer Science journal or in an LNBIP Series book.