A relatively new company, Jinni has been building a vast data base of descriptive tags tied to tens of thousands of TV shows and movies, which go well beyond traditional metadata to provide users intuitive means of searching for content and organizing personal libraries. The company now has relationships with major software providers, including NDS, OpenTV and SeaChange International, and is in “varying stages of engagement with operators in the U.S. and abroad,” according to Pohl.
The importance of advanced video search capabilities to cable operators’ efforts to create value for customers was underscored at the recent CableLabs Winter Conference in Denver, where attendees participating in an informal poll chose Jinni over ten other competitors for bragging rights in the organization’s annual Innovations Showcase. Other applications vying for operators’ votes touched on support for multi-screen convergence (SeaChange International), multi-room DVR support for content sharing (Pace Americas), ways to use mobile devices in TV applications (Trailer Park/Jargon Technologies), advanced home gateway support technology (Zenverge, Inc.), Web app-to-TV integration (Clearleap) and many other important categories.
Jinni’s appeal rests on the platform’s “Taste Engine” approach to recommending content, where users either choose from a menu of descriptions describing moods they’re in and types of content they feel like viewing or input descriptions in their own words. The engine goes through all the descriptive tags it has aggregated around content to find movies and TV shows that come closest to fitting the viewer’s description of what they’re looking for.
“Our Taste Engine technology takes browse, search and recommendations and combines the three into a tool that allows you to create a semantic profile of what you like and dislike in terms of content,” Pohl says. “You like witty off-beat comedies from the ’70s. I like gangster movies and pre-World War 2 one-man army shoot-’em-ups. Jinni will develop results and rank them so that the movie or TV show that is most likely to match what you’re looking for is at the top.”
The software, employing algorithms and interpretive language techniques, crawls the Web to discover all the reviews and other things said about particular movies and shows, ingests that information and tags it with up to 2,200 separate descriptors. The system also taps traditional metadata from DVR content and VOD content to round out its information base.
“Our semantic approach provides a superior experience that’s better than anything like collaborative filtering or other statistical methods,” says Roi Ophir, vice president of
marketing and product at Jinni. “It’s a very seamless search and recommendation engine that operates across linear, VOD and Web content.”
The system, which is available to consumers at Jinni.com for use in managing Netflix accounts or organizing their Web content libraries, will initially be implemented by service providers, aggregators and programmers in the IP domain, Pohl says. The company’s one announced customer is a Web content aggregator, Quickflix, a Netflix “clone” based in Australia.
But it won’t be long before it’s also used with traditional TV content, Pohl says. One approach entails service providers using their Web portals to allow users to search through the content available on VOD. When the recommendations are presented the user can decide what to watch and make the selection through the electronic guide on the TV set.
The next step would be to incorporate the engine directly into TV guides. “We’re very close to running the application on TV today,” Pohl says. “There are a number of ways to do that. The eventual migration to the guide could be as an EBIF (Enhanced Television Binary Interchange Format) app.”
The TV-based version of Jinni would be operated via the remote control rather than a keyboard, which would provide users menus for choosing descriptions without the option to type in their own semantic choices on a keyboard. “We have a five-key approach which we’re demoing on our Web site as a TV simulator,” he says.
One MSO taking a hard look at Jinni is Charter Communications. “We’re looking at using it with our VOD libraries and at integrating with our next-generation UI (user interface),” says Charter CTO Marwan Fawaz. “It could also play on TV Everywhere implementations with the video portals that most operators have. Using this technology makes for easier search and personalization of content.”
Jinni is pursuing various business models depending on the application. Where tie-ins with the set-top UIs are involved, the company will charge a licensing fee, Pohl says. “We’re doing a revenue share model on ad-supported Web sites,” he adds. “And where Jinni might be used to refer customers to content at a point of sale, such as a video store, we’ll take a piece of the revenue.”