Meta, Facebook’s parent company, today announced a new tool called Sphere, which is built around mining the vast repository of information on the open web to provide a knowledge base for work on artificial intelligence and other systems.
The first user of Sphere is Wikipedia, which is using it to automatically scan entries and identify when citations in its entries are strongly supported or unsupported. The research team has opened up the source code for Sphere, which is currently based on 134 million public web pages.
The idea of using Sphere for Wikipedia was simple: the online encyclopedia has 6.5 million entries and adds an average of about 17,000 articles per month. The wiki concept behind this actually means that adding and editing content is crowdsourced, and while there is a team of editors overseeing it, it is a daunting task that grows with each passing day, not just because of its size, but also because of its mandate.
Meanwhile, the Wikimedia Foundation, which oversees Wikipedia, has been weighing new ways to use all that data. Last month, it announced a corporate tier and its first two commercial customers, Google and the Internet Archive, which use Wikipedia-based data for their own business interests, and will now have broader and more formal service agreements surrounding it.
In Meta’s case, the company continues to be weighed down by poor public perception, in part because some accuse it of enabling the free dissemination of misinformation and toxic ideas, so launching something like Sphere feels a bit like a PR campaign for Meta, and if it works, and it could be a useful tool, it shows that someone in the organization is trying to work in good faith.
Today’s announcement about Meta’s partnership with Wikipedia doesn’t mention Wikimedia Enterprises, but in general, adding more tools to Wikipedia to ensure that the content it has is verified and accurate would be something that potential customers of the enterprise service would want to know when considering paying for the service.
It is not clear that the deal is making Wikipedia a paying customer of Meta. Meta did note, however, that to train the Sphere model, it created a new dataset (WAFER) of 4 million Wikipedia citations, much more complex than those previously used for such research. And just five days ago, Meta announced that Wikipedia editors were also using a new AI-based language-translation tool it had built, so clearly, there’s a connection there.
