Rummble Labs, which specialises in recommendation technology, is rolling out a new product today that promises to offer online publishers a new way to recommend content to users based on their Facebook accounts.
The London-based startup is launching Social Predictions, a recommendation layer which sits on top of Facebook’s Open Graph. Normally, when users log into a website with their Facebook account, they may get content recommendations based on their Facebook ‘likes’. This isn’t always that helpful though when taken literally. While you may ‘like’ Barack Obama, that doesn’t necessarily mean that you want to read articles about him all the time.
Social Predictions uses semantic and sentiment analysis to use your Facebook data in a smarter way. So, if you ‘like’ Ferrari on Facebook, there will be a higher chance of articles related to sports cars in general being recommended to you – not just that one manufacturer. It’s smarter with strings of words too. While ‘liking’ Pearl Jam, for example may be interpreted by some recommendation engines as meaning you like spreading jam on your toast, Social Predictions can link it to the band of that name and recommend articles not just about Pearl Jam but related music too.
Additionally, Social Predictions can intelligently suggest friends from a reader’s Facebook profile whose interests match the content they’re currently viewing, helping them share content with people who are most likely to appreciate it. This is all done while respecting Facebook’s native privacy settings.
Rummble Labs’ Alex Housley tells us that the company is rolling out the technology first via a deal announced last week with publisher Trinity Mirror. This will see the Social Predictions applied to the vast archive of articles on Mirror.co.uk. Other publishers interested in using the service can contact the startup for details.