Wikidata continues to play a more central role on the internet by supplying digital assistants like Siri and Alex with facts. Amplifying Wikidata’s facts through these digital assistants has implications for discovering new ideas, answering questions, and representing history accurately. In a recent Wikidata course, this exact concern came up in the area of criminal justice.
Wrongful convictions have happened consistently throughout history. This is well-documented on Wikipedia and on projects like the National Registry of Exonerations. There is also the Innocence Project, whose mission is “is to free the staggering number of innocent people who remain incarcerated, and to bring reform to the system responsible for their unjust imprisonment.”
In our Wikidata courses we introduce participants to several aspects of the open data repository — how linked data works, how to express relationships between items, and also how to create new ways of structuring information. This last concept is extremely important not only to ensure the completeness of Wikidata for the millions that use digital assistants, but also to ensure content is accurate and able to be queried.
If you consider how foundational structuring information is to the usefulness of Wikidata, then discovering information that has yet to be structured is one of the most important parts of contributing to Wikidata. The way to express these relationships on Wikidata is through properties (“depicts,” “country,” and “population,” for example). In one of our recent courses, a participant discovered a missing property and took the initiative to create it.
Ken Irwin, a trained reference librarian, has a passion for criminal justice reform. While searching Wikidata during our beginner course, he noticed that there were only properties concerning conviction. By only having conviction data, any post-conviction data could not be represented on Wikidata. The Wikidata community, of course, has a process for creating new properties. Irwin proposed an “exonerated of” property, and editors began discussing ways to structure this kind of data. An interesting question about data modeling followed.
Wikidata editors revealed potential ways to model post-conviction data. An “exonerated of” property would cover some of this information, but what about documenting data about pardons, amended sentences, and extended sentences? There is also an ongoing debate as to whether this information should exist as a qualifier, modifying a conviction property (since you cannot have an exoneration without a conviction). Anther school of thought suggested that exoneration data should exist as its own property since that particular person would have been cleared of any conviction.
These kinds of discussions have a direct impact on queries — how to pull this information from Wikidata, and how to associate it with other criminal justice data (i.e., where does this fall in the spectrum of rendering judgement, etc.).
After a period of debate, this property was approved in January 2020. You can see how many items use this property by clicking this link. Once that page loads, click the blue “play button” in the lower left of your screen to run the query.
These query results embody the word “wrong” in the phrase “wrongfully convicted,” and have a far reaching implication when it comes to describing people accurately. The ability to improve accuracy around the representation of information is one of the many reasons why so many people are drawn to working on Wikidata.
Stories like this underscore the importance of editors pursuing their passions, uncovering gaps, and taking steps to address those gaps on Wikidata. It is only in this way that those who choose to get involved in open data will be able to make Wikidata more reliable, equitable, and useful for all users and for anyone represented on Wikidata.
Interested in taking a course like the one Ken took? Visit learn.wikiedu.org to see current course offerings.