K-CAP 2025

Knowledge has played a foundational role in Artificial Intelligence since its inception. As AI systems become increasingly sophisticated, the demands on how knowledge is captured, represented, and operationalized have become more complex. Today’s breakthroughs—particularly in large language models and neuro-symbolic architectures—highlight the importance of moving beyond data volume toward precise, efficient, and meaningful knowledge representations.

K-CAP 2025 continues its tradition of convening an interdisciplinary community committed to advancing the science and engineering of knowledge capture. From structured semantics and multimodal fusion to learning-based extraction and human-in-the-loop systems, K-CAP offers a forum for exploring rigorous, creative, and scalable approaches to modeling and applying knowledge.

In a world increasingly driven by diverse, distributed data—from text and tables to images, video, and user interactions—this year’s conference emphasizes interpretability, compositionality, and reuse as key pillars of knowledge-centric AI. New to K-CAP 2025 is a strategic alignment with FAIR principles, promoting structured sharing of datasets, tools, ontologies, and methods to enhance transparency, reproducibility, and community impact.

The conference will feature a rich program of keynote speakers, tutorials, and workshops, alongside a vibrant poster and demo session. We invite researchers and practitioners from a wide range of fields—including knowledge representation, machine learning, human-computer interaction, and ethics in AI—to share their insights and innovations. K-CAP 2025 is committed to fostering an inclusive environment that encourages collaboration and the exchange of ideas across disciplines.

K-CAP 2025 invites participation from across AI research areas, including but not limited to: Knowledge representation and acquisition, Ethics and trust in knowledge-centric systems, Explainability and knowledge-driven interpretability, Interactive and intelligent user interfaces, Querying and reasoning over heterogeneous knowledge bases, Evaluation methodologies for knowledge systems, Neuro-symbolic and hybrid AI approaches, Compact models: constraint networks, graphical models, Knowledge in multi-agent and planning systems, Information and metadata extraction, Multimodal knowledge capture (text, images, tables, audio, video), Knowledge graph construction from language models, Structured knowledge extraction from LLMs, Deep learning for representation and reasoning, Representation learning from structured/unstructured data.

We look forward to welcoming you to K-CAP 2025—where symbolic and neural, theoretical and applied, human and machine perspectives converge to push the boundaries of knowledge capture.