Overview
Data-Centric Artificial Intelligence (DCAI) is revolutionizing the development and deployment of intelligent systems by shifting the focus from model-centric optimization to systematic data engineering. In the realm of intelligent agents, especially task-oriented bots, dialogue systems, and autonomous decision-makers, this perspective is particularly transformative. The ability of these agents to operate robustly and adaptively in dynamic, real-world environments depends critically on the quality, governance, and lifecycle management of the data they consume and generate.
DCAI4IA aims to bring together researchers, practitioners, and industry stakeholders to explore how DCAI principles can be applied to design, train, deploy, and monitor intelligent agents embedded in complex systems, services, and workflows. The workshop will serve as a forum for presenting new research, discussing emerging challenges, and showcasing applications where data quality and design directly impact agent behavior and performance.
This workshop is closely aligned with ICSOC's focus on intelligent services and workflows, particularly in how data-centric AI enhances the reliability and adaptability of service-oriented intelligent agents.
Workshop Objectives
- Highlight data-centric approaches that improve the performance, robustness, and reliability of intelligent agents.
- Promote exchange on methodologies for data management, enrichment, and governance in agent-based systems.
- Explore engineering challenges in integrating intelligent agents into complex, real-world processes and workflows.
- Discuss the impact of data practices on explainability, fairness, and adaptability in agent behavior.
- Share case studies and practical insights from industry and applied domains.
Topics of Interest
We welcome original contributions and position papers on topics including but not limited to:
Data Engineering for Intelligent Agents
- •Data preprocessing, cleaning, correction, and deduplication for agent training.
- •Dynamic data management and evolution in agent-driven environments.
- •Domain adaptation, data augmentation, and active learning strategies.
Data-Centric Models in Multi-Agent Systems
- •Data-driven coordination and decision-making in multi-agent settings.
- •Handling data drift and environmental changes in real-time.
- •Explainability and traceability via data provenance and metadata.
Generative AI and Conversational Agents
- •Dataset curation and prompt optimization for dialogue systems.
- •Evaluation frameworks centered on data quality and diversity.
- •Handling hallucinations and inconsistencies in LLM-based agents.
Architectures and Infrastructures
- •Workflow design and orchestration for autonomous agent ecosystems.
- •Monitoring and debugging data pipelines for deployed agents.
- •Integrating DCAI into MLOps pipelines tailored to agent lifecycle.
Applications and Case Studies
- •Intelligent agents in domains like finance, healthcare, education, robotics, logistics, smart cities, etc.
- •Comparisons of data-centric vs. model-centric designs in agent behavior.
- •Datasets, benchmarks, and reproducibility standards for data-centric agent development.
Important Dates
-
SEP30
Paper Submission Deadline
30 September 2025
-
OCT24
Notification of Acceptance
24 October 2025
-
NOV07
Camera-Ready Submission
07 November 2025
-
DEC01
Workshop Date
1 December 2025
Deadline Countdown
Submission Guidelines
We invite the submission of:
Position Papers
8-page limit
Present your viewpoint on workshop-related topics through thoughtful position statements. These contributions spark meaningful debates by posing critical questions, presenting alternative perspectives, identifying challenges, and proposing innovative solutions to emerging issues.
Research Papers
12-page limit
Share your cutting-edge research findings and comprehensive evaluations. Contributions may include work-in-progress studies, groundbreaking concepts with initial validation, state-of-the-art improvements, critical analysis of existing literature and practices, or practical experiences in theory implementation.
Industry Papers
12-page limit
Bridge the gap between industry and academia by showcasing innovative solutions that drive real-world impact. This track welcomes contributions that demonstrate strong industrial connections through comprehensive case studies, practical applications, or direct collaboration with industry partners as co-authors, fostering opportunities for funding and strategic partnerships.
Submissions must be formatted according to the ICSOC 2025 style guidelines (Springer's LNCS Formatting Guidelines) and submitted at: https://easychair.org/my/conference?conf=icsoc2025
Each submission will be peer-reviewed by at least three experts.
Target Audience
This workshop targets:
- Researchers in data-centric AI, multi-agent systems, and human-AI interaction.
- Developers of conversational and autonomous agents.
- Data engineers and MLOps professionals working with agent-based architectures.
- Practitioners deploying agents in domains requiring strong data handling (e.g., healthcare, finance, IoT).
Workshop Publication
Accepted papers will be published as post proceedings to appear in Springer together with the other workshops. A selection of high-quality contributions will be invited for extended versions in a special issue of a peer-reviewed journal.
Registration
The workshop will take place as part of ICSOC 2025, from December 1 to 4, with in-person attendance.
The register link will be announced soon.
Committee
Organizing Committee
- Nasredine Cheniki, University of Tours, France | Infeurope S.A., Luxembourg
- Abderrahmane Maaradi, University of Doha for Science and Technology, Qatar
- Sabrina Senatore, University of Salerno, Italy
- Yanmei Zhang, Central University of Finance and Economics, China
- Yacine Sam, University of Tours, France
Program Committee
- Antonio Luca Alfeo, eCampus University, Italy
- Sabine Barrat, University of Tours, France
- Auday Berro, École Supérieure d'Ingénieurs Léonard-de-Vinci, France
- Mohamed Reda Bouadjenek, Deakin University, Australia
- Omar Boucelma, Aix-Marseille University, France
- Mario G.C. A. Cimino, University of Pisa, Italy
- Yacine Gaci, Plus que PRO, France
- Juan Carlos González, University of Granada, Spain
- Amine Jaouadi, ECE Paris School of Engineering, France
- Nizar Messai, University of Tours, France
- Diego Rincon Yanez, Trinity College Dublin, Ireland
- Maria Angela Pellegrino, University of Salerno, Italy
- Giada Simonato, University of Pisa, Italy
- Mohamed Belaoued, Université de Reims, France
Organized by
Supported by
Contact Information
If you have any questions or doubts, please feel free to email us at:
dcai4ia@univ-tours.fr