CES 2012: Recommendation Engine Jinni Shops Around Interfaces
LAS VEGAS—The team behind semantic discovery engine Jinni think they have a better solution for creating video recommendations, and they’re at CES shopping for customers. Jinni has created reference designs to show how its recommendations could be used in an electronic program guide or by couch-surfers via a 10-foot interface.
Jinni offers recommendations based on mood and plot preference, so that a user can ask for an upbeat romantic comedy, for example, and get recommendations tailored to their mood at that moment. The engine is based on keywords created with an automated system that surveys plot synopses, reviews, and other writings about a film. It generates 40 to 60 keywords per title. In all, it uses over 2,200 natural language terms.
“You can look at Jinni as a kind of Pandora, but for movies and TV shows,” says Roi Ophir, vice president of product and marketing.
Jinni was recently licensed by Microsoft, but can’t say exactly how Microsoft is using its recommendations. It’s not able to mention any of its other licensees, says Ophir.
Jinni is also at CES showing off a new version of its website, one that will go live in mid-April. Users can tap Jinni’s recommendations in different ways: by looking for movies similar to one they like, by giving Jinni a look at their overall movie preferences, and by using getting recommendations from social networking friends with similar tastes. The site will soon offer recommendations for movies currently in theaters, Ophir says.
“We want to change the ways consumers consume entertainment,” adds Ophir.
TV and online viewers can now get recommendations based on their moods or favorite plot elements directly from Jinni.
Movie and TV recommendation engine now works with live programming; offers sandboxing for UI testing.
Speculation is that engine will power entertainment discovery on the Xbox, though Microsoft is mum.