D1.2 Data Management Plan

SAIFA aims to become Serbia's national hub for AI development, acting as an AI Factory Antenna linked to the Pharos AI Factory (Greece) and IT4LIA AI Factory (Italy). It provides a one-stop shop for computing access, AI tools, datasets, and consulting for academia, public administration, startups, and SMEs, built on national and partner computing resources (IPB's PARADOX and ITE's National AI Platform). Infrastructure resources and the access and use framework are described in the current draft of Deliverable D3.2, Report on infrastructure resources, access and use policies. The purpose of the Data Management Plan (DMP) is to contribute to good data handling through describing what data the project expects to generate and which parts of the data can be shared with the public. Furthermore, it gives instructions on naming conventions, metadata structure, storing of the project data and procedures to make public data available. The FAIR data management principles are described and their application in the case of SAIFA project, as well as a review of the resource allocation, data security and ethical aspects of data collection and usage. During the 36 active months of the project, Google Drive will be used for consortium administrative and management collaboration, including the output-tracking register. A partner-hosted Nextcloud-based solution is planned for controlled shared working and staging storage of research datasets and materials. Partner and HPC systems retain their processing and dataset-specific storage roles. Data will be handled in accordance with national and European rules on data protection and privacy.SAIFA will use an output-specific repository strategy to support its open-access and FAIR objectives. Zenodo remains the designated repository for public project deliverables, scientific publications, white papers and comparable public documents. A consortium-hosted DSpace repository is planned for eligible domain dataset releases, and a consortium-hosted GitLab instance for software, AI models, workflows, pipelines and associated technical artefacts. Publication is subject to rights, access restrictions and approval, with release timing described in Section 2. Public outputs will be linked from the project website and AI Gateway as appropriate. Publication repositories for other output types will be determined during implementation; a public dissemination classification does not by itself assign a repository. Published outputs will have appropriate identifiers, version information and descriptive metadata, including the project name and grant agreement number, implemented according to the relevant repository and output type. Creative Commons licences will regulate reuse of SAIFA's generated datasets, models, and publications, with software released under open-source licences where applicable. Data security arrangements will address administrative collaboration, controlled research staging, partner processing environments and the designated publication repositories. Published DSpace and GitLab outputs and their metadata will be retained after the project ends under the policies to be implemented in Section 6.3. Ethical aspects of data collection, generation and sharing have been considered. If the project processes special-category health data and data from minors; these datasets carry ethics questions to be resolved before use. Currently, one domain dataset is subject to third-party licence terms, which limit its reuse to non-commercial purposes. The questions on all affected datasets are handled under the oversight of the SAIFA Legal and Ethics Board. Nothing in this project shall be deemed to require a party to breach any mandatory statutory law, national or European regulation, or norm of research ethics. The DMP is a living document and will be updated throughout the project to reflect the actual research data generated. The final version of the DMP will be made available at the end of the project and will include instructions on how to access and reuse open datasets produced in SAIFA project. Day-to-day data management and monitoring will use an administrative output-tracking register in Google Drive, continuously updated to reflect actual data generation, working locations and published releases. Responsibility for maintaining the register will be assigned by the Project Management Board. This document is a result of the work on the WP1 and it present a set of guidelines for data collection, usage and storage. Together with other documents from the WP1, it should ensure that the main project objectives of this work package are achieved. Updates of this document will be provided and incorporated in the final version of this document which will be submitted as deliverable D1.2b (due date 36th month).