Subgroup 50 Data-format Requirement Document for an Automatically Readable, Comprehensive and Curated Experimental Reaction Database MEDUSAL

NEA/WKP(2026)4
Adobe Acrobat PDF Document - on 9/29/26 at 2:45 PM
- Adobe Acrobat PDF Document on 9/17/26 at 4:46 PM

This Working Paper constitutes the requirement document that guides the development of the experimental reaction database, Machine-readable Experimental Data User App & Library (MEDUSAL), created by the Nuclear Energy Agency’s (NEA) Working Party on International Nuclear Data Evaluation Co-operation (WPEC), Subgroup 50. Experimental reaction data are usually stored in the Nuclear Reaction Physics Database (EXFOR) in EXchange FORmat (EXFOR format). With MEDUSAL, the WPEC SG-50 wants to go beyond the EXFOR format and database to generate a library that is (a) automatically readable, (b) comprehensive and (c) curated.

To this end, a definition is provided here:
• How metadata and data should be described to be unambiguously interpretable by codes and algorithms;
• What information should be stored for each datatype;
• How the various stages of information and judgements on a dataset beyond the original publication can be preserved.

This information can range from updating a monitor cross-section to expert judgement applied to a data set for a specific evaluation. This database can serve as more automatically readable input for future evaluations, ML codes and preserves expert judgement of many man-years. 

Experimental nuclear reaction data underpin evaluated nuclear data used in areas as varied as criticality safety, neutron dosimetry, nuclear energy, non-proliferation and treaty verification. This Working Paper examines how these data are stored and used, and identifies a gap between traditional data compilation and the needs of modern, data intensive methods. It concludes that while the existing Nuclear Reaction Physics Database EXchange FORmat (EXFOR) remains essential, it is not sufficiently structured for systematic, automated use. This document therefore proposes a new approach that builds on EXFOR rather than replacing it, by reorganising and enriching the same data to improve consistency, transparency and usability across a wider range of applications.