Wikisource and Wikidata together: lessons from the Wikisource Conference

I have just attended the Wikisource Conference in Denpasar, Bali in the middle of February. Various topics are talked about regarding the betterment of Wikisource, but I think there’s one topic that piques my interest to the point of writing here. In a few presentations, I heard about the various methods of linking Wikidata and Wikisource, giving a great update on Wikisource’s capabilities. Here, I would like to discuss and propose how the Wikidata-Wikisource bonding can be.

Text annotation

For the start, let’s look at Wikisource first without concentrating too much on Wikidata. In one rather spontaneous presentation, he talked about a simple template called the Annotate QID. In short, the templates work by “annotating non-linked terms using Wikidata QIDs.”

Overall, it’s not a massive feat – it simply enables a brief description of an info to be showed when we hover the mouse to it, taken from Wikidata descriptions. Nevertheless, it’s a neat quality-of-life feature for computer reading (i.e. with mouse input), especially when it comes to certain occasions such as disambiguating a collection of same names, as featured in the template documentation. You can see it yourself in the documentation page. Be warned if you want to implement this in your Wikisource; this template uses a module called WikidataIB, which is large and critical, making it difficult if your Wikisource and/or your community doesn’t have the necessary expertise to localize it for your language.

Linking and formatting

Next, we go over to linking Wikisource material di Wikidata, and how the data formatting is done. This is presented by the Bangla Wikisource community represented by Bodhisattwa, who heavily references the Wikidata Books wikiproject for the formatting works in Wikidata, enabling a proper data handling of Wikisource materials. He detailed how Wikisource materials can be linked to Wikidata, and what potential are available, as done in their works in various Indian language Wikisource versions.

The full, detailed presentation is available in Commons, but I’ll put a couple of his matters over here.

Lexical referencing

One easy potential is lexical referencing. Wikisource houses a lot of olden dictionaries, spanning various writers, influences and periods, each determining how definitions were constructed for a single word/phrase (aka a “lexeme”).

This method is already done within the Wikidata lexicography team, where linkage manners to materials in Commons have be done. Take the lexeme entry for bunuh (L61383), Malay verb for “to kill”, where extensive definitions are scraped from various sources, including Dutch, French and Latin, especially from Commons material through the “gloss quote” (P8394) property. The usage example (P5831) data takes a Malay Wikisource example, specifically Syair Ken Tambuhan.

Quote referencing

In the presentation from my Malay Wikisource peer, Hadith Fajri, he talked about several findings we’ve uncovered during the first 10 months of Malay Wikisource, being launched in April 2024. In his presentation, he picked an example from our world’s magnum opus, Sulalatus Salatin or the Malay Annals, where there is a Persian quote being written. As such, he imagined a bonding between the Malay and Persian communities of Wikisource. However, I looked in another perspective specifically for this purpose – how can we link the Persian quote back to its origin?

Taking one step further from the linking works done in the previous section, now we can link between items to indicate quote referencing, using properties such as “quotes work” (P6166) and “quotation or excerpt” (P7081).

Another example Hadith showcased is the quoting of a hadith (in short, prophetic sayings in Islam) in both Sulalatus Salatin and a material from the Indonesian Wikisource. So, how can we do the linking? We can do the same as the one beforehand, but the nature here is a bit different – because hadith is a distinguishable primary religious material that deserves its own Wikidata item, much like for example, Biblical and Quranic verses. As such, instead of linking between manuscript items, we can link it to a hypothetical hadith item.

This opens a wild but interesting idea – what if we have items for all hadith in Wikidata, or how about every Biblical or Quranic verse? What fun things can we do with them? This is one question that I would love to ask myself.

Tech advancement

This is probably the most advanced potential I have seen so far in the conference regarding bilateral linkage. The Indian Wikisource community used the material’s Wikidata items to build a Wikisource reading app. This is possible in the ideal scenario depicted by them in their Meta page, where metadata is centrally stored in Wikidata items, enabling queries being done easily to build a list of compatible material, which are then processed through WIkidata API before export to the application front-end.

At the time I am writing this, the application is still in development, and as usual, discussions and improvements are much appreciated in my eyes. You can also looked at the app functioning in beta version here.

Conclusion

Wikidata has leveraged many data works through its openness. It’s time that we in Wikisource realize the full potential of Wikidata and integrate them together to fully explore the wisdom behind the olden works of humanity, especially. Wikisource is our free and open time travel machine to the past civilizations, where everyone can explore the wisdom of our forefathers, and I believe that Wikidata can gives us an extra hand to unlock the stories of the past, and I am eager to see how the community can maximize its potential during and afterwards the synergy. Indeed: “In the archive, we unite.”