Skip to content

Research Datasets

"Search_TERM_HERE" site:vision.in.tum.de OR site:www.cdbb.cam.ac.uk OR site:bimportal.scottishfuturestrust.org.uk OR site:digicatapult.org.uk OR site:pewresearch.org OR site:odsc.com OR site:archive.ics.uci.edu OR site:research.tudelft.nl OR site:archive.data.jhu.edu OR site:systems.jhu.edu
  • Global Deaths by Risk Factors Github Repo interactive spatio-temporal visualization of worldwide deaths related to various risk factors, specifically air pollution, substance use, and lack of sanitation.
URLCommentsFreeCategoryRegion
Smithsonian Library ResourcesDatabases, collections and search tools freely available via the InternetYAcademicGlobal
CrossSub (Alt)Micro-level, subnational event data on armed conflict and contentionYConflictGlobal
ACLEDReal-time data on political violence and protest events worldwideNConflictGlobal
OSMPOpen Source Munitions Portal by Airwars and Arms ResearchYConflictN/A
LiveUAFactual reporting on conflicts, human rights, protests, terrorism, etc.YConflictUkraine
Venezuelan Violence DataViolence statisticsYConflictLATAM
World City DatabaseCities with population and Lat/LongYCountry DataGlobal
TradingEconomicsMetrics and indicators by country over timeYCountry DataGlobal
GitNux Crime ReportsCrime reports and statsCrimeGlobal
Cloudflare RadarOutages, threats, rankings from Cloudflare dataYCyberGlobal
TUM DataComputer vision research datasetsYCyberGlobal
CEPAL Cyber AttacksCyber Attacks in LATAMYCyberLATAM
OECDForeign Direct Investment (FDI) StatisticsYFinanceGlobal
World Bank DataEconomic DatasetsYFinanceGlobal
Scottish Futures Trust ROIROI calculator for building projectsYFinanceUK
NumbeoCost of living calculator and comparison toolYFinanceGlobal
ReportlinkerAI enabled Market Intelligence PlatformYFinanceGlobal
KaggleData repository with datasets for competitionsYMLGlobal
YouTubeVideo database of Machine Learning contentYMLGlobal
AWS Open DatasetsVarious datasets from Amazon Web ServicesYDataGlobal
AMI Data SetData from finance and other sectorsYDataGlobal
Boston DataBoston city datasetsYCityBoston
Social EuropeSocial DataYSocialGlobal
US CensusCensus DataYDemographicsUSA
US Government DataGovernment DataYGovernmentUSA
UK Government DataGovernment DataYGovernmentUK
NewsData.ioAPI and Dashboard for news by countryYNewsGlobal
Chinese Military CompaniesNotice of Chinese military companies in the US (01/07/2025)YMilitaryChina

These small cards describe a specific collection, not a search portal. Publisher dataset and reuse documentation was checked for the cards; the validator checks structure, not data quality or permission to redistribute. Free access is not a license.

Collection and publisher sourceVersion/date and coverageUnit and limitationsReuse terms
World Development Indicators — World BankRevision not pinned; development indicators across economies, country groups and yearsEconomy/group–indicator–year observation; coverage and revisions vary by source and indicatorWorld Bank default CC BY 4.0 terms have additional conditions; third-party indicator metadata can override reuse permissions
ACS five-year estimates — US Census BureauSelected 2019–2023 estimates for reproducibility (later 5-year releases exist); US geographic aggregatesGeographic area estimate, not PUMS person rows; estimates have sampling uncertainty and varying geographic availabilityCensus Data API terms require attribution/non-endorsement and prohibit reidentification; they are not a blanket downstream redistribution license
Iris — UCI Machine Learning RepositoryDonated June 30, 1988; Fisher’s original study is from 1936; 150 plants and four featuresPlant sample; small classic benchmark, not broad population coveragePublisher identifies CC BY 4.0; attribution and other license conditions still apply

Machine-readable cards live in apps/wiki/scripts/data/dataset-cards/. From the monorepo root, use Python 3 (standard library only):

Terminal window
python3 apps/wiki/scripts/validate_dataset_cards.py apps/wiki/scripts/data/dataset-cards
python3 -m unittest discover -s apps/wiki/scripts/tests -p 'test_dataset_cards.py'

The CLI reads only selected JSON files/directories and writes JSON to stdout; it never downloads a dataset, checks a URL or gives a licensing opinion. Exit 0 means all selected cards validate, 1 means invalid/unreadable/no cards, and 2 is usage error. A file may contain one card or a nonempty array; every record is checked and duplicate IDs across the selection fail. Required fields are id, title, HTTP(S) source_url, owner, version_or_date, coverage, unit_of_analysis, nonempty limitations, calendar-date reviewed_on, and reuse_terms with status, url and explanatory note. Unknown terms use status: unknown, url: null, and a visible explanation. Synthetic positive and negative examples are separate in scripts/fixtures/datasets/. The optional record_count is allowed; unknown top-level fields such as an unreviewed license are rejected rather than treated as alternate reuse terms.

Recent changes

  1. fix: retire dead forum.irregularchat.com links across wiki, fieldnotes, bots, and docs gitayam
  2. feat(wiki): validate three public research dataset cards offline Hugo Zarate
  3. refactor(structure): flatten workers into apps/ directory gitayam