The SCIANCE Mapping Methodology provides a structured, scalable, and evidence-based approach to analysing the European landscape of Artificial Intelligence (AI) in Science.
The mapping approach combines semi-automated data analytics with expert validation and focuses on three interrelated dimensions: (i) key actors involved in AI-enabled scientific research, (ii) relevant research and innovation infrastructures, and (iii) emerging trends in AI adoption across scientific domains. The analysis thematically covers five scientific pilot domains—Astronomy & Physics, Materials Science, Earth Sciences, Life Sciences, and Social Sciences & Humanities—alongside cross-cutting AI research and innovations that support the scientific process.
At the core of the methodology is the OpenAIRE Graph, which serves as a trusted and interoperable backbone linking publications, datasets, software, projects, organisations, and services. A dedicated data pipeline enables systematic data collection, stratification, and enrichment, complemented by the automated detection of AI technologies in scientific texts, technology–application pairing, Technology Readiness Level estimation, and multi-dimensional trend and ecosystem analyses.
In addition, the deliverable defines the methodology for establishing the SCIANCE Good Practices Registry. Using a two-stage design—case identification and in-depth qualitative analysis based on a Context–Mechanism–Outcome framework—the registry documents empirically grounded, transferable examples of effective and responsible AI-enabled scientific practices.
Overall, the methodology supports SCIANCE’s strategic objective of strengthening Europe’s leadership in AI-enabled scientific research by underpinning the project’s Strategic Research and Innovation Agenda, infrastructure roadmap, and policy recommendations.
Disclaimer: This project deliverable has been submitted to the European Commission and is pending formal review and approval by HaDEA.
