Data Provenance

Release:

6.1.2

Date:

Oct 28, 2026

Data-provenance graphs recording the lineage of data artifacts in a set of Python projects, following the W3C PROV model (Entity, Activity, wasDerivedFrom) as studied in “Understanding Data Science Lifecycle Provenance via Graph Segmentation and Summarization”.

Graph	Num Nodes	Num Edges	Reachable pairs	Size (MB)	Download
sampleproject	148	763	342	0.004	sampleproject.tar.gz 📥
wikipedia-provenance	316	1284	1732	0.009	wikipedia-provenance.tar.gz 📥
pluggy	353	1818	880	0.008	pluggy.tar.gz 📥
itsdangerous	425	2210	2912	0.014	itsdangerous.tar.gz 📥
requests	682	3369	2524	0.016	requests.tar.gz 📥
httpx	826	4247	4408	0.011	httpx.tar.gz 📥
click	944	4654	2710	0.020	click.tar.gz 📥
jinja	1202	6277	7436	0.036	jinja.tar.gz 📥
flask	1517	7728	8062	0.041	flask.tar.gz 📥
fastapi	6455	26528	8814	0.094	fastapi.tar.gz 📥
celery	6880	35874	25306	0.158	celery.tar.gz 📥
scikit-learn	11706	60181	217728	0.725	scikit-learn.tar.gz 📥
sphinx	13431	72692	304084	0.974	sphinx.tar.gz 📥
pandas	13696	70959	154646	0.601	pandas.tar.gz 📥
django	23749	113137	615946	1.94	django.tar.gz 📥
zulip	59854	310032	209908	1.21	zulip.tar.gz 📥
superset	76348	394708	97460	1.03	superset.tar.gz 📥
airflow	90673	454738	91350	1.15	airflow.tar.gz 📥

Canonical grammars

The grammar file is attached to the archive.

\[\begin{split}\begin{aligned} S &\to Entity \quad Core \quad Entity \\ Core &\to wasDerivedFrom \\ &\mid wasDerivedFrom_{r} \\ &\mid specializationOf \\ &\mid specializationOf_{r} \\ &\mid wasGeneratedBy \quad Activity \quad used \\ &\mid used_{r} \quad Activity \quad wasGeneratedBy_{r} \\ &\mid wasGeneratedBy \quad Activity \quad Core \\ &\quad Activity \quad wasGeneratedBy_{r} \\ &\mid used_{r} \quad Activity \quad Core \\ &\quad Activity \quad used \\ &\mid wasDerivedFrom \quad Entity \quad Core \\ &\quad Entity \quad wasDerivedFrom_{r} \\ &\mid specializationOf \quad Entity \quad Core \\ &\quad Entity \quad specializationOf_{r} \end{aligned}\end{split}\]