141 treaties (1962–2025) · 9,870 pairwise comparisons · 38 families · 88 genealogy edges
DataverseNO DOI: 10.18710/WA7HEO · Dr. Oğuz Kaan Pehlivan, University of Oslo · CC BY 4.0
Five views generated only from the packaged Dataset 3 files. Click any figure to open it full size. Standalone PNGs are also provided, in figures/.
How similar treaty texts are across all 9,870 pairs, with the genealogy (≥0.45) and high-similarity thresholds marked.
The 141×141 cosine matrix reordered by family - bright square blocks are the textual families the clustering recovers.
The 38 families by size, coloured by generation profile; tied families are labelled and coloured explicitly.
Each treaty is linked to its most-similar predecessor in the deterministic (signature date, treaty ID) order; this also resolves same-day signatures.
Whether treaties that read alike (cosine) also share the same Dataset 2 clause coding - the two similarity notions largely agree.