AndreaBozzo/ceres-open-data-index

Ceres Open Data Index

A curated, deduplicated index of 349,836 open datasets from 25 CKAN portals across 9 countries plus international sources. 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 semantic search engine for open data, via the CKAN API. Metadata has been flattened from CKAN's nested JSON into a tabular schema, with noise filtered and cross-portal duplicates flagged.

What's included

• Dataset titles, descriptions, URLs, tags, organizations, and licenses from 25 CKAN portals
• Cross-portal duplicate detection (is_duplicate flag)
• Portal-level language annotations
• Per-portal Parquet subsets for targeted analysis

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)

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	True if the same title appears in another portal

Files

• all.parquet -- Complete dataset (349,836 rows)
• data/<portal>.parquet -- Per-portal subsets (25 files)
• metadata.json -- Snapshot metadata with counts

Splits by portal

Portal	Country	Datasets
data.gov.au	Australia	108,946
dati.gov.it	Italy	69,476
data.gov.ua	Ukraine	39,863
data.humdata.org	International	26,842
ckan.open.nrw.de	Germany	23,183
data.gov.ie	Ireland	20,632
ckan.opendata.swiss	Switzerland	12,780
dati.toscana.it	Italy	12,663
catalog.data.metro.tokyo.lg.jp	Japan	8,554
dati.regione.marche.it	Italy	5,427
data.gov.ro	Romania	4,880
dati.emilia-romagna.it	Italy	3,118
opendata.aragon.es	Spain	2,882
datos.gob.cl	Chile	2,803
dati.comune.milano.it	Italy	2,579
dati.puglia.it	Italy	1,782
dati.trentino.it	Italy	1,377
dati.regione.umbria.it	Italy	449
dati.lazio.it	Italy	403
dati.comune.roma.it	Italy	365
dati.regione.campania.it	Italy	331
dati.regione.sicilia.it	Italy	186
dati.comune.genova.it	Italy	158
dati.regione.liguria.it	Italy	124
dati.comune.napoli.it	Italy	33

Loading the dataset

import pandas as pd

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

# Load a specific portal
df_milano = pd.read_parquet("data/milano.parquet")

# Filter non-duplicates
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

1. Discovery: Each portal's CKAN API (/api/3/action/package_list) provides the full list of dataset IDs
2. Fetching: Dataset metadata is retrieved via /api/3/action/package_show with concurrent requests (10 parallel, circuit breaker for resilience)
3. Normalization: CKAN package JSON is normalized into a flat schema with title, description, URL, and content hash (SHA-256)
4. Incremental sync: Only new or modified datasets are re-fetched on subsequent runs, using content hashes and CKAN's package_search?fq=metadata_modified:[timestamp TO *]
5. Export: The ceres export --format parquet command streams all datasets from PostgreSQL, applies curation filters, flattens JSONB metadata, detects cross-portal duplicates, and writes compressed Parquet files

Curation

The following curation steps are applied during export:

• Noise filtering (6,532 rows removed): Titles shorter than 5 characters, titles containing "test"/"prova"/"esempio", datasets with empty descriptions
• Duplicate flagging (59,062 rows marked): Datasets whose title (case-insensitive) appears in more than one portal are flagged with is_duplicate=true but kept in the dataset
• Metadata flattening: Tags, organization, and license are extracted from nested CKAN JSON into top-level string columns

Source data

All data comes from publicly accessible CKAN API endpoints. No authentication is required. The portals are government-operated or institutional open data catalogs.

Update frequency

This snapshot was generated on 2026-02-25. The Ceres harvester can refresh the index at any time; new snapshots are published monthly.

Considerations

Known biases

• Geographic skew: Australia is the largest single portal (31%), followed by Italy (28% across all Italian portals) and Ukraine (~11%). Australia, Italy, and Ukraine account for ~70% of the index
• Language distribution: Predominantly English (Australia, Ireland, HDX) and Italian, with significant Ukrainian, German, Japanese, French, Spanish, and Romanian content
• Portal selection bias: Only CKAN-based portals are included. Major platforms using Socrata (US, UK), DCAT (EU), or custom APIs are not yet represented
• Aggregation duplicates: dati.gov.it aggregates regional Italian data, so ~17% of rows are flagged as cross-portal duplicates with regional portals

Limitations

• Metadata quality varies by portal -- some datasets have minimal descriptions or missing fields
• Tags and organization names are not standardized across portals
• The metadata_created and metadata_modified fields are portal-reported and may not reflect actual data updates
• License information is inconsistent; some portals use identifiers (cc-by), others use full titles (Creative Commons Attribution 4.0)

Ethical considerations

This dataset contains only public metadata (titles, descriptions, tags) from government open data portals. It does not contain personal data. All source portals publish data under open licenses (typically CC-BY, CC0, 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 (CC-BY, CC0, ODbL, or national open data 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 25 CKAN portals, snapshot 2026-02-25}
}

Links

• Ceres on GitHub
• CKAN API documentation