10 Years of Wikidata – Part 1

Andy Mabbett, aka Pigsonthewing, has been a well-known Wikipedian since 2003. He has contributed more than 940,000 edits across 162 projects in his 19 year tenure as a Wikimedia editor and was one of the very first contributors to edit Wikidata since its early days.

“Earlier than that, in about 2007 – he says – I did some work on ‘microformats’, which as a method of marking metadata within a page using HTML classes in order to specify that a piece of text was a date, a name, an address, geographical coordinates, or whatever. I did that mostly on external websites because I needed it for a project I was running for an organization. In 2007 I started to apply those markup techniques to English Wikipedia infoboxes so that computers could read that metadata.” In other words, Andy was doing some of the things Wikidata would do a few years later: “I didn’t know that at the same time people like Denny were planning a project to make metadata available through a database. But I guess we were both trying to do the same thing in different ways.”

Wikidata arrived in 2012, a clean slate of a knowledge graph where only Wikipedia sitelinks could be added for the first few months. “I got involved with Wikidata probably even before its official launch. At that time it was very basic and then, when we had the ability to add Properties, I contributed to some of the specifications of what the new Properties might be.” Working on Properties was, in fact, one of the various domains in which Andy contributed heavily: at least 175 Properties were directly proposed by him and many more proposals had comments and suggestions to make them better. “I’m not the only one that did that, of course, I worked with other people – he interjects – but I did my part to try to make sure that the project was robust, that it was going to survive because of its quality and it wasn’t going to get sidetracked. I think what we did in the early days was give it a good steer. We made some sensible ground rules and some of them are written down as policies now. Some others were just more a case of custom and practice, but we can now say to people ‘this is how you do it’. I suppose ‘developing best practices’ is the best way to put it.”

Meanwhile, as many other contributors did in their history of contribution, Andy started to populate Wikidata’s main namespace by adding Items about Wikipedia articles he had written and topics that were of interest to him. From Birmingham’s local historic buildings, to people he had written of, to some of the GLAMs (Galleries, libraries, archives and museums) he worked with. “I did this to get as much practice as I can on Wikidata, to see what it was all about. And then – he continues – I became involved with a number of bulk imports of data into Wikidata through my GLAM work. One of the first that springs to mind is the TED Talks. Jane Darnell and I were both recruited by TED as Wikimedians in Residence and we encouraged volunteers to write Wikipedia articles about speakers at the main TED conference since many of them didn’t have one. A lot of our work was also adding the speakers’ biographical data and details about their TED talks into Wikidata. This wasn’t any ‘scraping from Wikipedia’ initiative, this was one of the first in which gathering data was done by exporting the content from a database, reorganizing it and then uploading it in bulk through QuickStatements.”

Many other imports followed, from ORCID to BBC, from the British historical cultivars of apples, to Songkick, to Quora – which is probably one of the projects that Andy is more proud of, «because it was one of the biggest projects and it also had an impact on their website». Quora, in fact, started to draw data from Wikidata in order to improve its own ontology of data, more or less same as what Soundkick did as well. “The other thing I am particularly proud of was the dblp computer science bibliography import: I worked with them to import their identifiers and match them with Wikidata, which they imported into their own database, so that you have a two-way linking. By doing this, they were able to import other data from Wikidata and improve our project as well, by exchanging identifiers and data. I think this is our strength: we have a symbiotic relationship with our partner organizations that strengthens both in the long run.”

All in all, when asked about the next ten years of Wikidata, Andy feels there is still room for improvement: “there are some features I would like to see, maybe a gadget or a tool or a minor change to the core software, that would be nice to have. These do not represent a failure in Wikidata, the website is good and useful as it is, but these things would make it a little better.” Of course, there’s also the wish that more people understand and use Wikidata: “I am certainly a very strong advocate for the project. I think it’s a great pity that it is so badly underutilized on English Wikipedia; it could be so much better if it made more use of Wikidata, and if members of the community fed back their issues into Wikidata.”

Camillo Pellizzari, known on the Wikimedia projects as Epìdosis, is about to start his Ph.D. in Antiquity Sciences at the Scuola Normale Superiore di Pisa. He started contributing to the Italian Wikipedia in December 2012 and started to contribute to Wikidata in the following months.

“I immediately specialized in merging duplicate Items – he remembers – especially those about categories. I realized that the old interwiki system left out a lot of potential interconnections between projects, especially between languages that are not close to each other. At the time, there was no possibility to keep the second Item as a redirect to the first too, so I quickly amassed quite a lot of requests for deletion.” There were so many that Camillo decided just to propose himself as an administrator on Wikidata, “mostly to speed up the process of deletion”.

To date, according to the statistics, around 55,000 merges bear his signature, making Epìdosis the eighth contributor in that field. Camillo still continues to clean-up duplicate Items, but his focus shifted with the years from the duplicates that come from the projects to those that come from external database imports. “I started to grasp this ‘new’ aspect of Wikidata, in other words that Wikidata no longer was a project serving only Wikipedia, especially since 2019. Until then I mostly did imports of Wikipedia data into Wikidata or merging Items. I must say that sometimes – he concedes – I unfortunately still have some doubts about the quality of certain data imports, even though it really depends much on the import and on the topic.”

Camillo still holds true to the ‘old Wikidata values,’ “maybe – he quips – because of my background as a history student. I still explain, when I do a presentation about Wikidata, how the project started as a way to centralize data coming from various Wikipedias, in order to be reused on even more Wikipedias.” This, however, doesn’t prevent him from finding that the ten-year-long evolution of Wikidata has been extremely positive for the project: “Wikidata certainly evolved with time and I did too. I have expanded my work towards the interaction with external databases, especially GLAM databases, that want to transpose data from Wikidata in their own database and the other way round. It seems to me that the basic function of the project didn’t change, but it merely expanded its potential area of intervention to other entities which do the same thing that Wikipedia does: choosing which relevant data to show to their readers.”

Speaking of GLAMs, right before the start of the pandemic Camillo met in person with another contributor he casually got to know on Wikidata – a meeting that resulted in the birth of a brand new GLAM project in Italy: “I met Stefano Bargioni – User:Bargioni on Wikidata – because of his edits on Wikidata. In real life, he’s Deputy Director of the Library of the Pontifical University of the Holy Cross and he started editing Wikidata by adding the identifiers from his own library. I casually stumbled on his edits and we started messaging. In January 2020, we met in person in Rome, and I showed him Mix‘n’Match and other Magnus tools. Stefano immediately understood the potential and from there our collaboration took off. He then contacted other colleagues, who in turn realized the value of collaborating with us – Camillo continues – and so the ‘Gruppo MAB’ was created” (“MAB” is the Italian acronym for museums, archives and libraries). Almost all of the activities of the group were held online in recent years, “not just because of the various lockdowns, but also because we’re all in different cities. Nonetheless, we managed to have some in-person meetings and we plan to have others in the next months.”

Regarding the future Camillo wishes mostly two things. First: a way to make tool development and technical changes more easy to request. “It’s disappointing that there is no established process to request a gadget or tool – he notes – and have it realized by someone. Then we have the problem of community-maintained tools that sometimes accumulate a number of requests and bugs that are never addressed or very slowly fulfilled.” Secondly, an improvement on the so-called ‘data round-tripping’: “To me it’s paramount to forge close partnerships with all institutions that give us, or can give us, data, who can then act as data quality controllers. At the moment there is almost a complete lack of a workflow that would allow corrections in both directions, especially from Wikidata back to the original database, which is an almost non-existent feature. In my opinion, correcting someone’s data is as important as giving the correction back to its original source, since the quality of the other databases is key to maintaining the quality of Wikidata itself.”