Session date and time: Breakout Session 2 – 14:30-16:00 BST, Tuesday, 6 October 2026

Collaborative session notes

The primary focus of this session is to advance the core deliverables of the FAIR4ML Interest Group by reviewing current progress, engaging in hands-on collaborative refinement, and establishing a strategic roadmap for our future activities. Specifically, the discussion will be focused on the following objectives:

• Foster Knowledge Exchange: Provide an overview of ongoing initiatives within the broader ecosystem, including updates from NFDI4DataScience, the ELIXIR AI Ecosystem, and recent advancements in AI validation repositories.
• Continue the work around the ML Lifecycle for FAIR4ML (TF1): Evaluate the mapping of FAIR principles across the various stages of the machine learning lifecycle. Participants will engage in an interactive session to align current findings with the NFDI framework, establishing a clear understanding of challenges and opportunities for practitioners.
• Continue the effort around standardizing ML Model Metadata (TF2): Review the proposed metadata schema based on schema.org. The session will include an interactive workshop to define the foundational steps toward an ML Management Plan, drawing on successful models from the RDA FAIR4RS WG and ELIXIR.

Ultimately, the outcome of this meeting will be a set of concrete actions for the next 6-12 months, including community spaces (e.g., building a community of practice), for information sharing about ML and FAIR pertaining to ML.

FAIR4ML will actively pursue the identification and engagement with additional relevant groups.

• Welcome, Introduction to the FAIR4ML IG and how to join (5’)
• Updates from ongoing related projects (5’ each, 15’)
  • NFDI4DataScience and MLentory (Leyla Jael Castro)
  • ELIXIR AI Ecosystem / OSAI (Fotis Psomopoulos)
  • AI Validation linking to repositories (Johan van Soest)
• Output of TF 1: Presentation and discussion on the ML Lifecycle for FAIR4ML (30’)
  • Overview of activities so far (5’)
  • Interactive session (20’):
    • work between the mapping of the stages and the NFDI document.
  • Reporting (5’)
• Output of TF2: Presentation and discussion of a ML model metadata based on schema.org (30’)
  • Overview of produced output (5’)
  • Interactive session (20’):
    • Steps towards an ML management Plan, following the past example of the RDA FAIR4RS/ELIXIR activity
  • Reporting (5’)
• Possible new activities (5’)
• Next actions and wrap-up  (5’)

Researchers and data professionals interested in developing, deploying, sharing, and/or supporting Machine Learning solutions, focusing on how the FAIR principles can be interpreted in and applied to the context of ML to improve the development, deployment, sharing, and use of such models.

Particularly relevant are members of relevant RDA groups with complementary focus, in order to identify potential synergies. An initial list of these key groups is:

• Software Source Code Interest Group, in part as the maintenance home for the FAIR for Research Software Working Group (FAIR4RS) outputs
• FAIR Digital Object Fabric Interest Group
• FAIR Data Maturity Model Working Group
• Reproducibility Interest Group
• Artificial Intelligence and Data Visitation (AIDV) WG