July 1, 2009 weblog
Jinni: Semantic Search for Movies
(PhysOrg.com) -- One of the most interesting things I have come across over the Internet is the movie search engine Jinni. Can't think of anything to watch tonight? Type in a phrase, and dozens of moves appear in an attractive visual search result. Of course, if you want to see movies that match your tastes and preferences, you will have to set up a profile and become a part of the The Movie Genome. Jinni has been called "the Pandora of music" for good reason.
Because of the ability to compare tastes to others and rank movies according to your preference, Jinni seems like a social network at first glance. However, the people behind Jinni make it clear that it is an "Internet application designed to fit how people relate to movies and TV." As with all applications of this nature, the more you use it, the more accurate your results will become over time. I compared Jinni to the recommendation I get from Netflix, and found that Jinni offers something much more sophisticated. There's a reason that CNET thinks that Jinni is hands down the best movie recommendation engine available.
The real cool feature, though, is the semantic search. Type in searches according to plots, genre and/or actors. You can even search using your mood as a guide. You can type in nearly any search phrase and something will come up. If you are confused about the connections between some of the movies that appear, you can click on a button that will tell why the movies are "related". Jinni uses a proprietary algorithms to do a lot of the work, but the staff has done a great deal of manual tagging. Automatic tagging continues as Jinni "learns" more about different movies.
My favorite thing about Jinni, though, is the fact that it reminds me of old favorites not thought of in years -- and introduces me to interesting new movie possibilities.
© 2009 PhysOrg.com