How It Works

From raw records to linked data

The Publicspending.net team focuses on engineering meaningful interconnections among public spending data collected from government sources. Rather than leaving figures as disconnected spreadsheets, the project models payments, payers, and payees so they can be compared across jurisdictions and time periods.

The three-step process

Harmonizing companies and categories

Public spending records rarely use consistent names or category codes. The team develops algorithms to unify different names referring to the same company and to harmonize variform payment category classifications, including CPV and NAICS codes, so that data from different jurisdictions can be meaningfully compared.

At its core, the Publicspending.net project exists to make government expenditure legible. Across countries like the United States, Greece, Australia, and the United Kingdom, public agencies publish scattered records in different formats, classifications, and languages. The team behind this platform gathers those raw datasets and applies a consistent ontology to them, so that payments from different jurisdictions can be compared on a like-for-like basis. The result is a global view of who is paying whom, turning thousands of individual records into a meaningful, searchable picture of where public money flows.

The technical backbone of the platform is built around linked data. Rather than storing expenditure figures as isolated rows in spreadsheets, the project models payments as interconnected entities, linking payers, payees, companies, and categories through a unified structure. Classification schemes such as CPV and NAICS are referenced in parallel, enriching each record with additional context. This approach allows queries to trace relationships across datasets, revealing patterns and interconnections that would otherwise remain hidden in disconnected government files.

For users who want to dig deeper than the visual summaries, the platform offers multiple ways to engage with the data. Bulk datasets can be downloaded and reused for independent research, while a SPARQL endpoint allows technical users to execute custom queries against the underlying data store. The website itself presents tables, graphs, and statistics that summarize spending by category, location, and time. Whether a visitor is a journalist, a researcher, or simply a curious citizen, the goal is to make the trillions of dollars, pounds, and euros in public spending easier to follow.

The project is an ongoing engineering effort, with the team continually refining algorithms to unify company names, resolve entity variations, and improve classification accuracy across records. New datasets are added over time, and the platform evolves in response to feedback from the community of users who rely on it. While the site currently offers a beta version of its global dataset collection, the long-term vision is to expand coverage and deepen the level of detail available. Public spending affects everyone, and this initiative is built on the belief that the data behind it should be transparent, accessible, and genuinely useful.