AndreaBozzo/ceres-open-data-index

Ceres Open Data Index

A curated index of 2,564,009 open datasets from 419 open data portals across 40+ countries plus international sources (EU Open Data Portal, HDX, Copernicus/NASA/ESA geospatial catalogs), with cross-portal duplicates flagged. This is the largest aggregated open data metadata index available as a single downloadable resource.

Dataset Description

This dataset contains normalized metadata from government and institutional open data portals. It was harvested by Ceres, a harvest-first toolkit for open data, across nine portal families: CKAN, DCAT-AP (udata REST), SPARQL-backed DCAT, Project Open Data (data.json), Socrata, OpenDataSoft, ArcGIS Hub, OGC CSW, and collection-level STAC. Metadata has been flattened into a tabular schema, with noise filtered and cross-portal duplicates flagged (flagged, not removed).

What's included

• Dataset titles, descriptions, URLs, tags, organizations, and licenses from 419 open data portals
• A heuristic cross-portal duplicate signal (is_duplicate flag) — a title match, not canonical deduplication (see Curation)
• Portal-level language annotations
• Per-portal Parquet subsets for targeted analysis
• A versioned snapshot manifest, coverage/quality report, and snapshot-to-snapshot changelog

What's NOT included

• The actual data files (CSVs, shapefiles, etc.) -- only metadata
• Embedding vectors (model-specific; regenerate with any embedding model)
• Resource-level metadata (individual file URLs within datasets)

Note: SPARQL-sourced rows (e.g. data.europa.eu) carry title, description, identifier, and modification date; portal-specific fields such as tags/organization/license are populated mainly for CKAN/udata/Socrata/OpenDataSoft sources. STAC and OGC CSW rows are catalog/collection-level records, not individual files.

Dataset Structure

Schema

Column	Type	Nullable	Description
original_id	string	no	Dataset ID from the source portal
source_portal	string	no	Portal base URL
portal_name	string	no	Human-readable portal name
url	string	no	Direct URL to the dataset page
title	string	no	Dataset title
description	string	yes	Dataset description
tags	string	yes	Comma-separated tag names
organization	string	yes	Publishing organization
license	string	yes	License title or identifier
metadata_created	string	yes	Original creation date (ISO 8601)
metadata_modified	string	yes	Last modification date (ISO 8601)
first_seen_at	string	no	When Ceres first indexed this dataset (RFC 3339)
language	string	yes	Primary language code (e.g. it, en, de)
is_duplicate	boolean	no	Heuristic signal: the same title (case-insensitive) appears on another portal. Not canonical deduplication

Files

• all.parquet -- Complete dataset (2,564,009 rows)
• data/<portal>.parquet -- Per-portal subsets (416 files; names derived from the portal hostname, e.g. data/catalog-data-gov.parquet; rows repeat those in all.parquet)
• identity.parquet -- Slim per-record fingerprint (source_portal, original_id, content_hash), one row per all.parquet row, used to diff snapshots
• metadata.json -- Versioned snapshot manifest: stable snapshot_id, generation time, curation counts, per-file SHA-256 checksums, per-portal inclusion, and duplicate_detection provenance
• reports.json / report.md -- Coverage and quality report (coverage by portal/type/profile/language; field-completeness rates)
• changelog.json / changelog.md -- What changed versus the previous snapshot (added/changed/removed/unchanged, per portal)

Splits by portal

Top 48 portals by dataset count (of 419 total). Per-portal file names are derived from the portal hostname (e.g. data/catalog-data-gov.parquet, data/data-europa-eu.parquet).

Portal	Country	Type	Datasets
data.europa.eu	EU (international)	DCAT (SPARQL)	689,589
catalog.data.gov	USA	CKAN	399,597
www.data.qld.gov.au	Australia	CKAN	181,970
govdata.de	Germany	CKAN	158,177
data.gov.au	Australia	CKAN	139,729
data.opendatasoft.com	International	OpenDataSoft	91,100
dati.gov.it	Italy	CKAN	65,077
ckan.publishing.service.gov.uk	United Kingdom	CKAN	57,244
data.gov.uk	United Kingdom	CKAN	57,060
datos.gob.es	Spain	DCAT (SPARQL)	48,797
open.canada.ca	Canada	CKAN	47,437
data.gov.ua	Ukraine	CKAN	36,960
webstat.banque-france.fr	France	OpenDataSoft	35,837
data.gov.cz	Czechia	DCAT (SPARQL)	32,605
data.humdata.org	International	CKAN	28,039
data.gouv.fr	France	DCAT (udata REST)	23,000
healthdata.gov	USA	Project Open Data	22,417
data.gov.gr	Greece	CKAN	21,900
data.gov.ie	Ireland	CKAN	21,267
data.cnra.ca.gov	USA (California)	CKAN	21,114
data.overheid.nl	Netherlands	CKAN	20,796
dados.gov.pt	Portugal	CKAN	20,528
idee.es (CSW)	Spain	OGC CSW	18,690
pndb.opendatasoft.com	France	OpenDataSoft	17,710
edx.netl.doe.gov	USA	CKAN	17,186
ckan.opendata.swiss	Switzerland	CKAN	15,863
ckan.open.nrw.de	Germany	CKAN	14,910
data.e-gov.go.jp	Japan	CKAN	12,276
dati.toscana.it	Italy	CKAN	12,214
data.slovensko.sk	Slovakia	DCAT (SPARQL)	11,491
podatki.gov.si	Slovenia	CKAN	8,956
data.norge.no	Norway	DCAT (SPARQL)	8,638
catalog.data.metro.tokyo.lg.jp	Japan	CKAN	8,483
datos.gov.co	Colombia	Socrata	7,773
emodnet.ec.europa.eu (CSW)	International (marine)	OGC CSW	7,104
datosabiertos.gob.pa	Panama	CKAN	5,607
discover.data.vic.gov.au	Australia (Victoria)	CKAN	5,578
dati.regione.marche.it	Italy	CKAN	5,450
data.gov.ro	Romania	CKAN	5,018
data.virginia.gov	USA (Virginia)	Project Open Data	4,825
data.ca.gov	USA (California)	Project Open Data	4,506
catalogue.cioos.ca	Canada	CKAN	3,484
data.gov.rs	Serbia	DCAT (udata REST)	3,449
catalogue.data.gov.bc.ca	Canada (BC)	CKAN	3,343
justice.gov	USA	Project Open Data	3,240
datos.gob.cl	Chile	CKAN	3,037
geoportal.statistics.gov.uk	United Kingdom	ArcGIS Hub	3,024
data.ontario.ca	Canada (Ontario)	CKAN	2,910

...plus 371 more portals (~110,000 datasets), including US state/city Socrata and ArcGIS Hub catalogs, European OpenDataSoft portals, national geospatial CSW catalogs, and Earth-observation STAC APIs. The complete per-portal breakdown is in reports.json / metadata.json.

Coverage & quality

Figures below are generated with the snapshot (reports.json / report.md).

By source type: CKAN 1,445,993 · DCAT-AP 868,468 (SPARQL 793,235 · Project Open Data static_json 46,094 · udata REST 29,139) · OpenDataSoft 165,032 · OGC CSW 27,264 · ArcGIS Hub 20,818 · Socrata 5,074 · STAC 3,125 · unmapped 28,235.

By language (primary code per portal):

Language	Datasets		Language	Datasets
English (en)	1,865,304		Slovak (sk)	11,491
German (de)	192,890		Slovenian (sl)	8,956
Italian (it)	94,245		Norwegian (nb)	8,638
French (fr)	91,162		Finnish (fi)	5,086
Spanish (es)	82,553		Romanian (ro)	5,018
Ukrainian (uk)	36,960		Serbian (sr)	3,449
Czech (cs)	33,112		Croatian (hr)	1,755
Dutch (nl)	22,740		Latvian (lv)	1,534
Japanese (ja)	22,707		Polish (pl)	1,380
Greek (el)	21,900		Kyrgyz (ky)	1,313
Portuguese (pt)	21,387		Hebrew (he)	1,171

Plus Danish (da) 614, Swedish (sv) 383, Indonesian (id) 26, and 28,235 rows from portals not yet mapped in the export config (language/type = unknown; mostly small geospatial catalogs). language is the portal's primary language, not per-row language detection.

Field completeness (share of rows with a non-empty value): description 100.0% · modification_date 55.6% · organization 54.3% · tags 48.1% · license 33.7%.

Loading the dataset

import pandas as pd

# Load all datasets
df = pd.read_parquet("all.parquet")

# Load a specific portal (US federal)
df_us = pd.read_parquet("data/catalog-data-gov.parquet")

# Drop rows flagged by the heuristic duplicate signal (title match, not canonical)
unique = df[~df["is_duplicate"]]

Or with the datasets library:

from datasets import load_dataset

ds = load_dataset("AndreaBozzo/ceres-open-data-index")

Dataset Creation

Harvesting methodology

Ceres harvests each portal through the API best suited to it, with a streaming, resilient pipeline:

1. CKAN portals are paged via package_search (streaming, content-hash delta detection). The client adapts on slow/flaky portals: per-request timeout grows and page size shrinks on timeouts/truncated bodies, then recovers, so large catalogs page fully without manual tuning.
2. US data.gov is harvested via its relocated action API on api.gsa.gov (API-key authenticated); datasets stay attributed to catalog.data.gov.
3. DCAT-AP (udata REST) portals (e.g. France, Serbia, Luxembourg) are paged via the catalog JSON-LD endpoint, with rate-limit cooldown and retry on transient non-JSON pages.
4. SPARQL-backed DCAT catalogs (e.g. data.europa.eu, datos.gob.es, data.slovensko.sk) are paged with LIMIT/OFFSET queries and deduplicated by dataset URI.
5. Project Open Data (data.json) static catalogs (US federal/state agencies) are read as a single DCAT-US document.
6. Socrata portals are harvested from the Discovery API, scoped to the portal domain and limited to dataset assets.
7. OpenDataSoft portals use the Explore API v2.1; catalogs deeper than the API's 10,000-row window are walked with a keyset cursor on modified.
8. ArcGIS Hub portals use the Hub Search API; hosted-service and file items are normalized to their Hub landing page.
9. OGC CSW catalogs are harvested capability-first (CSW 2.0.2 / ISO 19139), and STAC APIs are indexed at Collection granularity (never individual Items).
10. Export: ceres export --format parquet streams all datasets from PostgreSQL, applies curation filters, flattens metadata, detects cross-portal duplicates, and writes compressed Parquet.

data.europa.eu remains a partial SPARQL harvest: rows with English or untagged titles are represented (~690k of a portal reporting ~1.98M), while datasets titled only in other languages remain out of scope.

Curation

The following curation steps are applied during export:

• Noise filtering (53,061 rows removed): titles shorter than 5 characters, titles containing "test"/"prova"/"esempio", datasets with empty descriptions
• Duplicate flagging (1,192,910 rows marked): datasets whose title (case-insensitive) appears on more than one portal are flagged with is_duplicate=true but kept in the dataset. This is a heuristic signal, not canonical deduplication — it performs no entity resolution. The matching rule and its version are recorded in metadata.json under duplicate_detection (title-exact-ci-cross-portal, v1). Known mirrored/alternate endpoints can be declared as portal aliases so they are not counted as independent sources. Treat the flag as "likely duplicate by title", not a guarantee.
• Metadata flattening: tags, organization, and license are extracted from nested source JSON into top-level string columns

Source data

All data comes from publicly accessible open data catalogs operated by governments or institutions (CKAN, DCAT-AP, SPARQL, Project Open Data, Socrata, OpenDataSoft, ArcGIS Hub, OGC CSW, and STAC endpoints). Only the US data.gov source requires an API key (free).

Update frequency

This snapshot (ceres-20260715-ed0c345e8663) was generated on 2026-07-15. It adds 709,414 datasets and updates 121,302 versus the previous snapshot (ceres-20260626-87b9b74e080d), with 1 removed and 1,733,293 unchanged (see changelog.md). The Ceres harvester can refresh the index at any time; each refresh publishes a new manifest, report, and changelog.

Considerations

Known biases

• Geographic skew: the EU Open Data Portal (data.europa.eu) is the largest single source (27%), followed by the USA (catalog.data.gov ~16%, plus many state/city Socrata, ArcGIS, and data.json catalogs) and Australia (13%, dominated by Queensland and the federal portal). Germany, the UK, Italy, Spain, Canada, and France each contribute large national catalogs; many smaller regional/municipal portals round out the tail.
• data.europa.eu coverage: the snapshot contains 689,589 rows from a portal reporting roughly 1.98M datasets. Datasets titled only in languages other than English are under-represented.
• Language distribution: predominantly English (~73%), followed by German, Italian, French, Spanish, Ukrainian, Czech, and 18 more languages. language is the portal's primary language, not per-row language detection.
• Aggregation duplicates: national and cross-portal aggregators (e.g. dati.gov.it, data.europa.eu, the OpenDataSoft federation hub) re-list datasets from regional/member portals, so ~47% of rows are flagged as cross-portal title duplicates.
• Geospatial tilt in new sources: much of the growth since the previous snapshot comes from geospatial families (ArcGIS Hub, OGC CSW, STAC) and large statistical/energy catalogs, which skews new coverage toward geodata and scientific metadata.

Limitations

• Metadata quality varies by portal -- some datasets have minimal descriptions or missing fields
• Tags and organization names are not standardized across portals
• metadata_created/metadata_modified are portal-reported and may not reflect actual data updates
• License information is inconsistent across portals (present for ~34% of rows)
• SPARQL, STAC, and OGC CSW rows carry fewer populated columns than CKAN/udata/Socrata rows
• ~28k rows come from portals not yet mapped in the export config and carry type/language = unknown

Ethical considerations

This dataset contains only public metadata (titles, descriptions, tags) from government and institutional open data portals. It does not contain personal data. All source portals publish data under open licenses (typically CC-BY, CC0, ODbL, or equivalent).

Additional Information

License

This dataset is released under Apache 2.0, consistent with the Ceres project. The underlying metadata is sourced from portals that publish under open licenses.

Citation

@misc{ceres-open-data-index-2026,
  title={Ceres Open Data Index},
  author={Andrea Bozzo},
  year={2026},
  url={https://github.com/AndreaBozzo/Ceres},
  note={Aggregated metadata from 419 open data portals, snapshot 2026-07-15}
}

Links

• Ceres on GitHub
• CKAN API documentation