WikiCite panel: Should we split scholarly articles out of the Wikidata graph?
2021-10-31, 15:30–16:00 (UTC), Room 3

Bibliographic metadata is currently a large part of Wikidata. Statements about scholarly articles make up 50% of the triples in Wikidata, and are heavily used by tools like Scholia, but are only touched by 2% of all queries.

Over the past months an analysis was begun of the possibility of splitting scholarly articles out from the Wikidata graph, including the size of the subgraph and potential performance impact of the change. https://phabricator.wikimedia.org/T281854

We will discuss the ongoing results of that analysis, when it makes sense to consider splits at all, what a split might look like (how would it happen, how would queries be federated, how would this impact tools like Scholia?), and how we might reach decisions about such changes.

Panelists: Lydia Pintscher, Daniel Mietchen and Aisha Khatun; Moderator Sam Klein


What will the participants take away from this session?

A better understanding of what splitting scholarly articles from Wikidata would mean in practice.

Language

English

Recording

Yes

Link to notes

https://etherpad.wikimedia.org/p/WikidataCon2021-WikiCitepaneldiscussion