Looking for something to watch tonight? Jinni, a new app still in private beta, recommends movies, shorts and TV shows that fit your personal tastes.  Sort of like Pandora for movies, Jinni’s team is working on a Movie Genome Project to classify movies and TV shows. When you give Jinni information about movies you like and movies you dislike, Jinni’s “taste engine” uses the information in the to index to bring back a customized list of recommendations. You can also search for movies and narrow your search by genre, plot, mood and more.  Like Pandora, Jinni’s recommendations improve over time.

To learn more about Jinni, I interviewed Yosi Glick, Jinni’s co-founder and president, via email:

1. How did you come up with the idea for Jinni?

“While methods for getting movies / video into the hands of viewers have greatly advanced, methods for helping viewers figure out what they actually want to watch are lagging. In other words, we’re overwhelmed with choice.”

“Genres (e.g. action, comedy), rather vague and stuffy categories originally designed for back-office cataloging, are still the favorite way of presenting options. In my decade of working in the TV industry, I saw how limited data prevents TV operators from designing more intuitive interfaces for their video-on-demand (VOD) offerings.”

“Jinni takes an entirely new approach to choosing what to watch. We call our search-and-recommendation engine a taste engine, because the focus is on understanding and answering to people’s personal tastes. Semantic search on Jinni is a flexible, mood-based approach to exploring, while recommendations are based on a unique model of each person’s taste.”

2. Can you explain how the Movie Genome project works?

“Our Movie Genome contains several thousand genes, assigned to each title to describe plot, mood, style, setting, soundtrack and more. The Movie Genome is our core intellectual property and powers both search and recommendations on Jinni.”

“The Movie Genome taxonomy was designed by film professionals. New titles are automatically indexed via analysis of user reviews and metadata, using an innovative Natural Language Processing solution This automated process offers consistency, a diversity of viewpoints (from analyzing many reviews) and of course efficiency.”

“Users on Jinni can vote on whether a given gene fits a given title, and we’re planning to add more features for people to participate in the Movie Genome Project.”

3. About how many movies/tv shows have you indexed so far?

“To date we’ve indexed about 20,000 titles – and going strong!”

4. What makes the recommendations people get from Jinni different or better than recommendations from sites like Netflix and Amazon.com?

“Jinni takes a fundamentally different approach to recommendations. While Netflix and Amazon use collaborative filtering methods (based on the familiar “People who liked this also liked – ” approach), Jinni creates a unique model of each person’s taste based on descriptors from our Movie Genome. This approach has many advantages, including recommending lesser-known titles (because we’re not relying on other people to have watched and liked them first), explaining in detail why a recommendation has been made, and adjusting recommendations based on mood and other factors. You’ll see the recommendations page on Jinni evolving in this direction.”

5. What made you choose the name Jinni?

“Jinni is actually an alternative spelling for “genie.” We chose it because Jinni helps you find whatever your heart desires (when it comes to movies and shows), plus we like how it sounds.”

Pretty cool, but how well does it work?  Jinni actually does do a good job of suggesting titles you’ll like.  For example, a search for movies similar to “Pan’s Labyrinth”  came up with a page of recommendations. Some were old favorites, others were more obscure movies I’d never seen before. A search for movies like “Lord of the Rings” brought up the usual suspects (Eragon, the D & D movie, etc), more obscure fantasy flicks, some anime movies  AND  the old live-action He-Man movie I’d completely forgotten existed but now have a crazy desire to see again. Ah, the 80′s!

The user interface is extremely well-designed, and makes browsing through movie and TV show titles a pleasure.  Once you find something you like,  the “Watch it” option takes you to a page where you can rent it, buy it or stream it via i-Tunes. Or, link Jinni to your Netflix account, and you can have the movie deposited into your queue. Now, if only I had a working DVD player…

Jinni’s currently in private beta. You can get on the mailing list here, OR try to snag one of the 9 beta invites I have left. Here’s how:

Follow @usefultools on Twitter and re-tweet this post. The first 9 people to follow and retweet this post get beta invites for Jinni.

Ready, set…GO!

Our Rating of Jinni

Tags: Lifestyle

Categories:web