There are more than 600,000 profiles recorded in EveryPolitician, our database of politicians, officeholders, and lawmakers - but there’s a lot more work to do.
In a world of near-constant political flux, where politicians are voted in and out, installed, or, in some cases, overthrown, maintaining a global, up-to-date database of politicians is a task that’s far bigger than any engineering team.
With the help of contributors around the world, we’re hoping to fill in some of the gaps in our political data and build a dataset that can be used to hold governments to account and empower citizens.
Expanding and deepening our data means improving our tooling: alongside EveryPolitician, we’ve released PoliLoom, a crowdsourcing tool designed to fill in missing data about politicians and simplify contributions to Wikidata.
Credit: Caroline Ommer via Unsplash
Why is political data so fractured?
Our Politically Exposed Persons (PEP) data relies heavily on Wikidata, the structured-data sister of Wikipedia, which contains over 120 million entities. This data repository is an invaluable source, built on the work of hundreds of thousands of contributors worldwide. However, its multilingual, collaborative nature makes it both messy and vast.
There are also many technical and political nuances to political data in Wikidata. Different countries have different political systems and terminology, with roles such as “Minister” or “Member of Parliament” recorded in different ways.
Then there’s the question of how to record officeholders in disputed territories and de facto governments, and the issue of systemic bias: while well-known politicians in countries such as the U.S. may have detailed entries, others in smaller or less digitally documented nations often have incomplete entries.
All of the above mean that gaps are common: politicians are often missing birth dates, positions, or other basic identifiers or context. That information often exists in Wikipedia articles but isn't yet structured in Wikidata, and making it machine-readable requires significant manpower — manual data entry doesn't quite scale to hundreds of thousands of political positions worldwide.
Better tooling, improved data
One way we can address these gaps is by making it easier for individuals to contribute and verify political data. PoliLoom breaks down common data collection tasks, such as adding birthdates or position dates, into bite-sized, five-minute contributions that fit easily into a quick coffee break.
The tool processes the complete Wikidata dump (100+ million entities) and uses Large Language Models (LLMs) to extract politician data from Wikipedia articles and match it to Wikidata entities that are missing that information.
But human verification is still a vital step in the workflow. Using PoliLoom, contributors can compare the extracted information and the source document side-by-side and accept — or reject — the extracted details.
Verified data will not only feed our PEP dataset in EveryPolitician, but will be available to Wikipedia and thousands of other projects powered by Wikidata.
An example of a politician profile in PoliLoom, with the extracted details on the left, and the source on the right.
Filling in the gaps
EveryPolitician is a reimagination of the original project created by UK non-profit mySociety, which was paused in 2019 due to the challenge of sustaining this kind of work when funded by grants.
As the organisation pointed out in their 2019 blog post, politician data “is always steadily going out of date”. Political cabinets are constantly shifting, politicians come and go, and governments change. They also noted that changing governments often come with revised websites and data formats, sometimes breaking scrapers in the process.
While we appreciate that EveryPolitician will never be complete — and is far from perfect — we’re working to make it more consistent, and creating the right tooling is the first step in this process. Our hope is that PoliLoom will help bridge the gap between scattered information and structured data, all while bringing a little fun to the process.
Come and help us track who’s running the world
Log in to PoliLoom with your Wikidata account, or sign up if you don’t have an account. You can select which languages you want to work in, and there’s a tutorial that will walk you through the basics.
And, while you’re here, we’re running a PoliLoom collaborative demo on 2 April 2026, which you can add to your calendar here or join via this link on the day. We’ll take you through the tool and why it matters, and there’ll be an opportunity to ask questions. If you need anything in the meantime, get in touch via the OpenSanctions Forum or GitHub!
