Methodology and sources

This atlas shows every Marie Skłodowska-Curie project from 1996 to 2026. The Commission's own dashboard starts in 2014, and the anniversary map presents 88 selected stories. This is the whole corpus of 42,441 projects, so how they were selected matters.

Where the data comes from

Everything comes from the open data of CORDIS, the EU’s official database of funded research. The snapshot used here is dated 2026-09-13. A pipeline re-downloads and processes it, recording the URL, date and checksum of every file.

How Marie Skłodowska-Curie projects are identified

The programme was renamed twice and sat in a different branch of every framework programme, so there is no single code to filter on. The rules are:

The judgement calls

• Research training networks are included. In FP5 (HPRN) and FP4 (FMRX) they were not yet called Marie Curie, but in FP6 the same instrument officially becomes Marie Curie RTN. Leaving them out would break the timeline. They are flagged as included after review.
• Scheme codes are not used for FP4. 142 FP4-ESPRIT projects carry EIF, OIF, IIF and EST codes that mean Marie Curie from FP6 onwards, but there they are IT projects. FP4 therefore relies on the programme branch alone.
• Access to large facilities is excluded. FMGE in FP4 and HPRI in FP5 fund infrastructure, not researcher fellowships.

How precise the locations are

A project has no single place. The map shows the coordinator and every partner, with lines between them. CORDIS supplies coordinates for most organisations; the rest are placed from an offline gazetteer by city, and failing that at the centre of the country. Approximate points are drawn hollow so they never look more precise than they are.

Differences from the official figures

• The MSCA dashboard counts from 2014 and reports more than 18,000 projects. For the same period this atlas has 20,425.
• The Commission reports more than 150,000 supported researchers. CORDIS does not publish fellow names, so this atlas counts projects and organisations, not people.
• Figures shift with each monthly CORDIS refresh and as new grants are signed.

The texts

Project descriptions are the original CORDIS texts, in English. They are always on the project page.

Above them, in Bulgarian, sits a two-sentence explanation written by a language model (gemma4:cloud) from the CORDIS text — 40,107 projects so far. Each one carries an "AI summary" badge and sits in its own section, apart from the original. The model sees only the project description: it does not search the web and adds no outside facts. Its instructions forbid marketing words and any promise the source does not make.

Before anything is published it passes an automatic check: text outside 12–70 words is rejected, so is text that is not in Cyrillic, text containing forbidden words, and any word that welds the two alphabets together. What is rejected is simply not shown, and the page keeps only the CORDIS original.

These summaries are an aid, not a source. Where they differ, the CORDIS text and the link to the project are what counts.

Licences

• Project data: CORDIS, © European Union, reuse decision
• Map tiles: OpenFreeMap / OpenMapTiles, © OpenStreetMap
• Geocoding: GeoNames (CC BY 4.0)

Found a mistake?

Write to admin@nauka.bg. Every project page links to its original CORDIS record, so anything here can be checked.

Who makes this atlas

The atlas is made by Forum Science Association (BG Science) as part of K-TRIO — Researchers in the Knowledge Triangle, the European Researchers' Night project in Bulgaria, coordinated by Sofia University St Kliment Ohridski. The programme for the Night on 25 and 26 September 2026 is at night.nauka.bg, and the project itself is described in this article.