Wikipedia may be the go-to resource on almost everything these days, but according to Meta, it’s filled with dodgy, inaccurate citations.
But don’t worry, the company says its AI is here to help, having developed Sphere, a model capable of automatically scanning hundreds of thousands of citations at once to check whether they truly support the corresponding claims.
Meta claims it created a new dataset of 134 million public web pages as a knowledge source for the model, which says is “an order of magnitude larger and significantly more intricate than ever used for this sort of research”.
Sphere (opens in new tab) uses open web data rather than traditional, proprietary search engines such as Google, and has already compiled 134 million documents from across the web.
Built using CCNet (opens in new tab), a variant of Common Crawl, Meta says Sphere will help other AI researchers working on knowledge retrieval projects.
Meta says the eventual goal of the project is to build a platform to help Wikipedia editors systematically spot citation issues and quickly fix the citation or correct the content of the corresponding article at scale.
The tool reportedly calls attention to questionable citations, allowing human editors to evaluate the cases most likely to be flawed without having to sift through thousands of properly cited statements.
If a citation seems irrelevant, Meta says its model will suggest a more applicable source, even pointing to the specific passage that supports the claim.
The news comes as Wikipedia is looking towards new ways of raising revenue other than donations.
The Wikimedia Enterprise (opens in new tab) platform recently announced it would start to charge companies such as Google, Amazon, and Facebook which use Wikipedia as a resource.
You can grab the source code for the project on GitHub here (opens in new tab), and interested parties can also read a full write-up of the project’s findings here (opens in new tab) or access the demo here (opens in new tab).