Last Saturday (August 19th) at Newspeak House in London, mySociety and Wikimedia UK held the “Wikifying Westminster” workshop, a day-long event to encourage people to get involved with Wikidata, but also to give a taste of what people can build with the data that is already there.
The vision: one day, complex investigations which currently take researchers a lot of time, such as “how many MPs are descended from people who were also MPs” or “how many people named X were MPs in year Y”, will be answerable with data from Wikidata using a single SPARQL query…
…but we’re not quite there yet. Currently, some data is scattered all over separate databases (which sometimes get shut down or disappear); some is just plain missing; and most frustrating of all, some is in place but there’s no apparent way to get it out of the database.
In order to make this vision a reality, we need to experiment with the data, find ways to check how complete it is, and explore what questions we can currently answer with it. Events like Wikifying Westminster are the perfect opportunity to do just that.
After a brief introduction to Wikidata and the EveryPolitician project, we split into two groups: one focused on learning how to use Wikidata, while the other focused on working on mini-projects.
Here’s a taste of what happened…
The learning track began by introducing new users to the basic Wikidata editing principles (or “getting data into Wikidata”). Participants were able to put their new skills into action immediately, by adding missing data on British MPs, who were mostly lacking dates and places of birth.
By the end of the first session, good progress had been made, particularly on obtaining dates of birth for current British parliamentarians. For some reason, though, it proved much harder to find these for women than for men: we can only speculate as to why that might be (do some still adhere to the idea that a woman shouldn’t reveal her age?!).
We were also given an introduction to SPARQL, a language used to query information on databases (or “getting data out of Wikidata”). Lucas Werkmeister introduced the Wikidata query service and explained a few tricks to help with using it. Participants were later able to put this to the test by running progressively difficult test queries such as “All current UK MPs” or “Who is the youngest current MP?”
Also, Navino Evans showed us the potential of reusing data, talking about Histropedia, which he co-created with Sean McBirnie. Histropedia is an awesome tool that lets you visualise thousands of topics on interactive timelines: you can browse through existing ones or create a new one from scratch.
This group both worked on improving data and looked at how well we could answer some simple “stepping stone” queries (i.e. small questions to which we already knew some of the answers) as a heuristic of how good the data in Wikidata already is. You can see and contribute questions to the list of test queries here.
Some more details:
Improving data. The focus here was on the Northern Ireland Assembly, for which Wikidata now has full membership history back to the foundation of the Assembly, and on adding academic degrees of cabinet ministers. Starting from an excellent spreadsheet of the undergraduate universities and subjects of UK politicians and ministers (going back to John Major’s cabinets), we tried to upload that data on the relevant items, adding the qualifier “academic major” (P812) to the property “educated at” (P69). In this case, the key problem we found was that we weren’t sure how to model when people did joint subjects, like “Maths and Politics”, convincing us to concentrate on the more obvious subjects first.
Answering some unusual and/or intriguing questions. Inspired by a prior finding that there are more FTSE 100 CEOs named John than there are female ones, and that John is historically the most common name of UK parliamentarians, we thought we’d find out when exactly the John-to-female balance was toppled amongst the UK’s MPs (hint: not until 1992).
Going back further in history, we queried the first time each given name was recorded in Parliament, this was inspired by a recent news article about an MP who claimed he was the first “Darren” in the Commons.
Some ideas were also born that we weren’t able to see through, for various reasons. For example, could we discover which, if any, MPs are descended from people listed in the UCL’s ‘Legacy of British Slave-owners’ database? An interesting question, but at the moment, the answer is ‘no’, partly because child-parent relationships are currently inconsistently modelled in Wikidata, and partly because of the nature of Wikidata and ancestry: if there is someone who doesn’t exist in Wikidata (e.g. Grandad Bob, the painter) in the family chain, Wikidata can’t bridge the gap between a present day MP and the slave owner who might be their ancestor.
This is just the beginning
Work, of course, is still ongoing: all pre-1997 UK data is still to be inserted or improved on Wikidata, and so much more is missing – family connections, academic degrees, links to other databases, and all sorts of “unusual stuff” that can be used for interesting queries.
This data is crucial if we want to be able to answer the really big questions which Wikidata should one day be capable of helping us explore, about what politicians do.
We can do that together!
We hope that events like this give people an easy way in to Wikidata and also show them what’s already possible to achieve with the data. Over the coming months, we are hoping to support more events of this type around the world. If you are interested in getting involved, here’s how:
- Want to improve your country’s data? Events like this can be a great way to help kickstart activities and find other people who share your goals. We are happy to help out and support people in other countries to do so.
- Are you already organising or planning to organise a similar workshop around Wikidata? Make sure it is listed on the Wikidata Event page!
- Do you want to attend future workshops? Follow us on Twitter to stay updated about events that we are running, and ones that other people are too!
We’re also always looking for feedback and suggestions on workshop and event formats that might also work. Have you already run similar workshops? Let us know your impressions and suggestions on firstname.lastname@example.org!
Feature image credits: Mark Longair
This post is by Tony Bowden and Lucy Chambers from the EveryPolitician team. Today we officially launch our collaboration with Wikidata – here’s what to expect…
The story so far
You might have been following the progress: since 2015, through our project EveryPolitician, we’ve been gathering data for every national legislature in the world, from thousands of sources, and sharing it.
Now, two years on, we’ve started to see some great results. For example:
- It’s much easier to build simple Parliament Tracking sites (as mySociety partner organisations have in Zimbabwe and Nigeria). Those running the sites can work on providing information and context to hold politicians accountable — and don’t have to worry about wrangling data and software.
- Tools that allow citizens to write to their representatives (like Majlis Nameh in Iran, based on the WriteInPublic software, or Oxfam’s UK and Australia Campaigning Tool) can now be deployed in days rather than weeks, allowing groups to focus on local customisation.
- Vote-tracking sites like TheyVoteForYou in Australia can be more easily adapted to other countries — such as Ukraine, again without worrying about the major task of sourcing the politician data.
- Projects that highlight politicians’ activity can be augmented to show extra information against those politicians — so for example, Politwoops can show party affiliation next to politicians’ tweets.
- Investigative journalists can cross-match our lists of politicians to other sources (e.g. investigating shell companies for a recent Private Eye article or with the Panama Papers).
- mySociety’s collaboration with Facebook makes it easy for people to connect with their newly elected representatives.
All of this is possible because of two main tenets of EveryPolitician: having gathered the data, we structure it consistently, and we share it freely.
Not just current data
We’ve also discovered that there is a huge value for everyone in retaining historic information, too:
- Old data tends to disappear from many official sites, for example, when a new government comes into power and / or when parliaments decide to remodel their websites.
- Sometimes even the official sources no longer exist: for example, in Burkina Faso the Parliament building was burned down by protesters in 2014 and many crucial documents were lost.
- Research projects and sites run by civil society organisations sometimes run out of funding and have to shut down — meaning that the data people were relying on can vanish overnight.
The future: going one step further
So that’s all great, but we think EveryPolitician can do still more to help the worldwide community of Civic Tech coders and activists.
In particular, we want to go beyond data showing who the politicians are, and also provide information on what they do. That’s because we can see real value in the ability to answer questions like:
- When we look at politicians who vote on issues such as gay marriage, smoking bans, tax on sugary drinks, etc, anywhere in the world — are there any broad correlations like age or gender?
- As countries elect more women MPs, do those women also gain equivalent representation in committees? Does this affect attendance and participation rates?
- Are there standard political career paths that can be observed anywhere in the world? So for example, do certain Cabinet positions limit future progression; and if so, would it be feasible to spot politicians who are on the way up, or on their way out?
- Do politicians who move from the lower to upper house act, and vote, differently?
- Do politicians change their voting activity after certain types of intervention, for example after receiving funding from oil companies?
… and there are undoubtedly millions more questions, each one just as interesting and with answers that could enrich our understanding of the world. We are aiming for a future in which each should be answerable within minutes, rather than form the basis of a multi-year post-graduate research project.
How we plan to get there
You may remember our recent proposal to integrate more deeply with Wikidata. We’re delighted to say that our proposal was accepted, and that makes EveryPolitician’s path very clear.
Wikipedia is fast becoming one of the best sources of political information in many countries: it’s often updated more quickly than major news outlets or official parliamentary sites are.
Which is great, but there’s an issue when it comes to using that information for projects like the ones we’ve mentioned above: Wikipedia contains largely unstructured information (that is to say, information that comes in a wide variety of different formats) — as you’d expect from any project with multiple contributors and, often, a free-text input.
There are also a lot of differences between Wikipedias in different countries. Some countries’ data (particularly countries where the Wikipedia community is larger) gets updated very quickly. Smaller language Wikipedias can’t rely on such a large pool of editors, and it tends to be longer before they are updated.
Additionally, and as you might expect, Wikipedia content will appear fastest in the countries most directly affected by the change being documented, so for example when there are elections in Estonia, the Estonian Wikipedia may show the results almost immediately, but it can take a while for those changes to trickle into all languages.
But of course, nothing’s ever quite that simple. While Wikipedia has a wealth of unstructured political information, on Wikidata there’s still an awful lot of data missing. You may recall that we recently ran a drive to ensure that every country at least had its head of government entered, but that’s just the beginning: in order to answer the kinds of questions we mention above, we really need to ensure there’s consistently-structured information for all legislators, all elections… and much more.
What you’ll see in the coming months
While we plan for EveryPolitician to retain its own identity and keep its own front end, we’re excited to say that over the next few months, we’ll be teaming up with Wikidata communities across the world.
Our first objective is see if we can bring Wikidata up to the same level as EveryPolitician for as many countries as possible.
And once we hit that target, we plan to go much further. The beauty of Wikidata is that you can add pretty much any information you want to add to politicians (indeed, to anything!), so local communities can decide for themselves which information is most pertinent and make sure that it’s included.
To help us make all of this happen, we’re expanding. If you’ve read this far and you’re still finding the project interesting, the chances are you’d be a great addition to the team! We’re looking for a new community co-ordinator, working from anywhere in the world (compatible timezones allowing): you can see all the details here.
How you can get involved
There are plenty of ways that you can help with this drive to improve the structured information on Wikidata. Here are the obvious ones:
- There are active missions always ongoing on the Wikiproject Heads of State and Government. There’s one there now – check them out!
- Would you be interested in organising a push to Improve Wikidata for your country? Get in touch on email@example.com and we’ll do what we can to help.
- If this all sounds interesting, but you’re not sure where to start, or you’re unfamiliar with Wikidata – drop us an email and tell us what you’re interested in. We’re more than happy to help you get started.