The integration of Explainable AI (XAI) with Knowledge Representation (KR) and Knowledge Graphs (KGs) has emerged as a critical field of study, addressing the growing need for AI systems that are transparent, explainable, interpretable, and trustworthy. Knowledge Representation offers methods for describing, organizing, encoding, and reasoning with domain-specific knowledge, providing a foundational layer for AI understanding. Knowledge Graphs, as a seminal result of KR, further enable the structuring of interconnected concepts and relationships, forming an intuitive framework for describing complex domains. KGs have been effectively used as a powerful tool for improving results in solving different tasks in multiple fields such as question answering, recommendation, medical decision support systems, fact-checking, semantic search, image classification, and many others. Together, KR and KGs provide a natural complement to XAI techniques, empowering AI models to produce meaningful explanations that align with real-world contexts and user expectations.
As AI systems become increasingly embedded in critical domains such as healthcare, finance, and law, the need for interpretable models that offer insights into their reasoning processes has never been more urgent. By combining XAI with KR and KGs, researchers and practitioners can develop systems that bridge the gap between technical outputs and human comprehension. This integration not only enhances the clarity and relevance of AI-generated explanations but also supports the development of fairer, more accountable systems by incorporating domain knowledge and logic into the reasoning process.

This workshop seeks to explore the rich opportunities and challenges at the intersection of XAI, KR, and KGs. Key themes include leveraging structured knowledge to enhance explainability and interpretability, using XAI methods to refine and validate KR and KG models, as well as to increase the trustworthiness of machine learning models, and applying these combined approaches to tackle real-world problems. We invite contributions on theoretical advancements, innovative tools, and case studies that demonstrate how knowledge-driven AI can deliver explanations that are transparent, domain-aware, and user-centric. By fostering collaboration among researchers, industry practitioners, and domain experts, this workshop aims to drive forward the development of ethical and impactful AI systems that align with human values and societal needs.

TOPICS

The main topics include but are not limited to:

Techniques and best practices for building interpretable machine learning models intertwined with Knowledge Representation (KR) and Knowledge Graphs (KGs).
Graph-based interpretability in neural networks and deep learning models.
Leveraging KR and KGs to infer causality and improve explanations in AI.
Enhancing reasoning capabilities in AI through symbolic and sub-symbolic Knowledge Representation (KR).
Combining traditional KR methods with modern XAI techniques for enhanced explainability.
Using ontologies and taxonomies to structure interpretable AI explanations.
Theoretical frameworks for explainability within KR and logic-based systems.
Explainable reasoning methods in rule-based and knowledge-based systems.
Explainability in dynamic and temporal Knowledge Graphs.
Factual and Counterfactual explanations for KGs.
Scalable approaches for real-time explainable reasoning using KR and KGs.
Evaluation protocols, metrics, and benchmarks for assessing the quality and clarity of KR- and KG-based explanations.
Cross-domain evaluation of XAI methods in knowledge-driven AI systems.
Interactive and adaptive explanation frameworks using KGs.
Personalization of explanations through contextual KR and user feedback.
Bias and fairness in explainable Knowledge Graphs and KR.
Identifying and mitigating systemic biases in AI explanations through KR.
Ensuring fairness in knowledge-driven XAI applications.
Hands-on tools and open-source libraries for implementing XAI and KGs in real-world settings.
Novel methodologies for integrating probabilistic KR with XAI.

SUBMISSIONS

Submissions must be written in English, prepared using the new CEUR-ART 1-column style (which you can download here, also available as an Overleaf template here), formatted in PDF, and submitted through the EasyChair workshop page https://easychair.org/conferences?conf=xaikrkgecai2025. XAI+KRKG 2025 invites submissions of research, industry, and application contributions. There are two submission formats:

Regular papers (max 12 pages excluding references): must contain enough substance that they can be cited in other publications and may not have appeared before.

Short papers (max 6 pages excluding references): results and ideas of interest to the XAI-KRKG audience, including position papers, system and application descriptions and presentations of preliminary results, and an overview of papers accepted at another conference. In the latter case, extended abstracts must clearly state the venue where the paper has been accepted.

All accepted papers are expected to be presented at the conference, and at least one author of each accepted paper must travel to the ECAI venue in person. Submissions should be single-blind so the names of the authors will be visible to the reviewers and should be indicated on the submitted files.

PROCEEDINGS AND POST-PROCEEDINGS

Meeting the criteria, proceedings of XAI-KRKG 2025 are planned to be published at CEUR Workshop Proceedings.

ORGANIZATION

Roberto Barile – University of Bari, Italy
Claudia d’Amato – University of Bari, Italy
Valeria Fionda – University of Calabria, Italy
Flavio Giorgi – Sapienza University of Rome, Italy
Antonio Ielo – University of Calabria, Italy
Ilaria Tiddi – Vrije Universiteit Amsterdam, The Netherlands
Gabriele Tolomei – Sapienza University of Rome, Italy