SPARQL Endpoint
Query the Publicspending.net linked open data directly using SPARQL, the standard query language for RDF datasets.
Query the dataset
Publicspending.net publishes its collected spending data as linked open data modeled with the Public Spending (PSNET) ontology. Alongside downloadable bulk datasets, a SPARQL endpoint lets researchers and developers run custom queries directly against the underlying triples.
Requests for endpoint access are handled through a request form, and example queries are provided to help new users get started with the query syntax and available data structures.
PREFIX psnet: <...>
SELECT ?payer ?payee ?amount
WHERE { ?payment psnet:payer ?payer ;
psnet:payee ?payee ;
psnet:amount ?amount . }
The SPARQL endpoint at Publicspending.net opens the door to the underlying linked open data that powers the entire platform. Instead of relying on pre-packaged downloads or visual dashboards alone, researchers and developers can issue direct queries against the triple store, asking precise questions about how public money flows across governments, agencies, and vendors. This capability transforms the site from a static reference into a living, queryable knowledge graph, letting users explore interconnections between payers, payees, companies, and spending categories in ways that predefined reports simply cannot match.
The Public Spending (PSNET) ontology provides the semantic backbone that makes endpoint queries meaningful. By modeling expenditure records with consistent entity classifications and references to standardized coding schemes, the dataset supports parallel analysis across different jurisdictions and time periods. Whether someone is tracking payments in Greece, the United Kingdom, Australia, or a state like Massachusetts, the unified data model allows comparisons that would otherwise require painstaking manual reconciliation. This careful attention to ontology design is what separates a mere collection of spreadsheets from a genuinely interoperable dataset.
Getting started with the endpoint is designed to be approachable even for those new to SPARQL. Example queries illustrate the core syntax, showing how to navigate the dataset's structure and retrieve meaningful results around spending categories, payees, and economic classifications. For those who need bulk access or more specialized engagement, a request process handles endpoint access in a structured way. The goal is to lower the barrier to entry so that journalists, civic technologists, and academic researchers alike can begin following the money without needing to reinvent query patterns from scratch.
The engineering team behind Publicspending.net focuses on developing algorithms that unify company names, resolve entity variations, and enrich expenditure records with parallel classifications such as CPV and NAICS codes. These efforts ensure that queries against the endpoint return results that are both complete and trustworthy, even when source data arrives in inconsistent formats. By continually refining these linkages, the platform strengthens the meaningful interconnections between datasets, making it possible to trace money from a government payer to a corporate payee across multiple levels of detail and jurisdictional boundaries.
Note: the sample query above illustrates the general shape of PSNET-modeled data; exact prefixes and predicates are defined in the ontology documentation referenced from the How It Works page.
Example query themes
Payments by jurisdiction
Retrieve payment records filtered by country or region, such as the United States, United Kingdom, Australia, Greece, or the tracked US states and cities.
Payer and payee lookups
Find payments made by a specific government payer or received by a given payee entity within the dataset.
Category breakdowns
Explore spending grouped by harmonized category classifications such as CPV and NAICS codes.
Time-bound queries
Constrain results to the date ranges covered by each jurisdiction's dataset, from the 1990s through 2013.
Learn more
See how the underlying data is gathered and modeled, review the research behind the entity-resolution and classification algorithms, or read about the team and initiative as a whole.