Movies are easier to access than ever, but many of us still struggle with the age old question, “What to watch?” The days of milling around Blockbuster may be gone for many people, but most cable boxes and online movie stores offer little in the way of movie recommendations. Jinni, a new startup launching today in private beta, is looking to help. The site has compiled an index of 10,000 movies and television shows that can be searched using natural language. TechCrunch readers can grab one of 500 exclusive invitations by registering here.
The site has created what it calls “The Movie Genome” - a database of movies tagged by a team of humans aided by a computer algorithm, with attributes spanning fifty categories. The database is reminiscent of Pandora’s Music Genome Project (which is also sorted by human professionals), though it is significantly smaller at this point. Users can either search for movies based on a manual search, browse through movies by their attributes, or can generate recommendations after completing a brief test that determines their movie personality. After finding a movie or show they’re interested in, users can buy or rent them (or in some cases, watch them free) through a number of linked services, including Hulu, Amazon, and Netflix.
In practice the search seems to work well. Each match is visually displayed in a grid as a thumbnail, with the most relevant matches emphasized with larger images. This style makes it easier to quickly identify movies you might be interested in, and also makes false matches less jarring.
Jinni will see heavy competition from existing movie sites like Netflix, which have invested years into developing accurate recommendation algorithms (Netflix even offers a $1 million Prize if anyone can best its algorithm). But Jinni has a fun, intuitive interface and seems to work well, so it may be able to carve out its own slice in the market. Other players in this space include Flixter, which offers social recommendations and TheFilter, which launched movie recommendations earlier this year.
See all
Post
The visual layout is great because most people want to watch “a movie like [insert name here] but with more action”. This could be a big help
Ummmm, wasn’t this called “Firefly” back in the Wild Web 1.0 days, using the hot technology called collaborative filtering? And didn’t they get shut down after getting bought by Microsoft, even before the dot com bubble burst?
All of this has happened before, and it will all happen again…
Depends who stands behind it, I get impression people at Jinni are really trying to insert more fringe cinema into their system , and look at that white house on the Washington meddows, one person color that or other is making a difference.
I’ve been using Jinni for a few weeks now (in the private alpha), and as a real movie buff that enjoys mainstream cinema, as well as indie flicks, I have to say I’m really impressed with this service.
I found more than a few movies I never knew even existed and already had the chance of viewing some of them - and most of the times, the recommendations were in place. I’ve tried some recommendations engines in the past, but they all seem to bring up the same old stuff I’m already familiar with, which is not the case with this service.
I really believe Jinni is a treat for movie fans.
Jinni has a pretty snazzy and fast web2.0 style interface. However while it purports to help you find stuff to watch, it seems more intent on gathering ratings from the user via a “taste-test” which has 10 categories. This seems kind of arbitrary and simplistic given the genome billing. Presently there’s nothing like an easily accessible ranking list which is what most users might find easiest to use. Also no ranking data seems to be pulled in from other prominent sites e.g. imdb. So it seems that jinni will depend on the “intelligence of crowds”. That’s good if you’re into crowds and their opinion. Most users don’t go to web sites to be “tested” rather to find information. Maybe others have ample time and can warm up to the jinni challenge.
Netflix’s algorithm maybe difficult to best it if using only computers to do so. But it’s algorithm is not that great in recommending movies. Human tagging is essential, and maybe Jinni can achieve that.
The most useful feature for me so far is clicking on ‘More Like it’ when looking at a particular film
ahh, you can just search for “like: something” - thats handy
I like it
Yay! netflix for none-members…
Haha, these guys have definitely visited TED before.
The Jinni collage, a maze of movie opportunities, is almost as much fun as the movies themselves. This is definitely a website you can use to visually enhance domestic arguments about movie choice.
I like the idea - it’s great that someone’s finally gotten around to making a site like this - but its still really rough. The collage is a cool concept, but I got bored with it after ten minutes. They have still yet to give me a reason to prefer them over Netflix (8 million users, including me) / Rotten Tomatoes / IMDB.
Looking forward to future iterations.
If you type in “anal”, the leftmost result is
No Entry (2005)
haha, good to know!
Might as well move this one straight to the deadpool.
Thanks for letting know about Jinni
I like the site. I’ve always have a hard time figuring out which movie to choose. Now that I’ve started using the site I enjoy searching for a new movie - or even TV show. The site is very helpful. I like the collage and it’s easy to see the plot and review as the mouse moves over the pictures. You can even find a movie similar to a TV show you like. I took the taste test and now get recommendations based on what I like. It’s as if I have my own personal advisor ready with recommendations. I have sound movies I never knew existed. I find this a fun and refreshing site.
Wow, excellent picks dude.
jess
http://www.anonymity.at.tc
TasteKid.com is also a great movie recommendation system. It works for books and music, too.
http://Movielens.org is pretty great alternatively, too.
in German: http://www.moviepilot.de - gives recommendations based on your taste. for theatres, german television program and DVD catalogue. Already out of beta with many of the features jinni is experimenting with. BUT in german.
At Jinni, our approach to search and recommendations is to take the analysis of people who’ve actually watched the movies, both our in-house content team and reviewers everywhere. So the results on Jinni for “Like: Blade Runner but faster” are relevant in plot, pace, style, etc. Compare this to collaborative filtering. Everyone knows the formula “People who liked this also liked…” - an essentially blind statistical correlation of people’s preferences that naturally tends to highlight what’s already popular, used by Amazon, Netflix, and others.
Similar to Pandora, recommendations on Jinni are based on analysis of the elements (“genes”) each person enjoys in movies. Right now we determine this through ratings, which is why we ask for ratings in the Taste Test and elsewhere on the site (we’re working on additional methods for collecting this info).
Gathering feedback is a top priority for our private beta test, so please try out Jinni and then let us know what you think!
This is awesome I will definitely check this out. Hopefully they have a good mix of international movies in the database.
Would love to see this as an iPhone app similar to urbanspoon. I could avoid many, many arguments with my husband over what to order on demand.
I’ll try it out but don’t Flixster and Spout et. al. already do this? There are far fewer movies than songs and I would think for most buffs there would be few surprises, especially people who tend to stick to single genres.`
There are a whole lot of movies out there, including indies that are much harder to discover. IMDB has around 1M titles and even if you’re able to dig out only the ones that fit your taste it’ll take you a few life times to see them all. Also, the time and money spent on a movie is much greater than on a song so good tools to find what you really want to watch are useful.
Due to popular demand, we’ve increased the number of invitations to Jinni available here, so Techcrunch readers are welcome to continue to sign up. And we appreciate feedback!
http://www.tastekid.com/ is also a great movie recommendation site. It works for books and music, too.
http://www.movies.monigo.com is pretty great, and working! alternatively, too.
I am prepared to really love this site but the user interface has one big weakness which will drive me away.
The ratings bar- it is just too hard to manipulate. The target is too small to mouse around the slider. It is just way too frustrating to rank things. Let me click on the bar and have the pointer go to the spot where I am pointing or use the arrow keys on the key board, or enter a number.
doesnt netflix already do that? i dont see how they could ever compete :/