After watching your share of this year’s Oscar nominees, you might find yourself following the time-honored tradition of renting every tangentially related film in the hopes of finding something else just as good. Type in Slumdog Millionaire or The Wrestler, and Netflix will offer up recommendations that are “more like this.” But if you are looking for quirkier recommendations, the kind that you’d find at your local independent video store, then head on over to Clerkdogs. The human-curated movie recommendation site has an OscarMatch feature that suggests 400 films similar to this year’s crop of Oscar nominees.
On Netflix, if you are looking for movies that are like The Wrestler, for instance, it comes up with Requiem For a Dream (same director), Slums of Beverly Hills, Sideways, and Sin City (also starred Mickey Rourke). On Clerkdogs, it comes up with a bunch of down-on-their-luck fighter films: Raging Bull, Requiem for a Heavyweight, and Fat City, a 1972 boxing movie described as “equally downbeat.” The Mickey Rourke pick is the classic Barfly, which seems more appropriate, and the sleeper pick is a 2008 movie called Ballast, a movie about a poor family in Mississippi described as having a “similar style” as The Wrestler. I doubt any computerized recommendation engine would make that match unless it saw an overlap in people who rented both movies.
Take a look at the other nominees and you’ll find the same type of serendipitous recommendations. For Slumdog Millionaire , you get Monsoon Wedding and Salaam Bombay! Those make more sense to me than Netflix’s suggestions of Little Miss Sunshine, Trainspotting, and Thank You For Smoking. I’m not sure Clerkdogs’ OscarMatch always comes up with better recommendations, but it does come up with ones you won’t find elsewhere.
Nice suggestion. I liked the name of site..very funny.
isn’t this the same functionality (or used to be) as http://www.spout.com?
Interesting. Is Netflix’s API for authentication only? Perhaps it would be better as a platform API that allowed users to supplant with competitive suggestion engines and still link back to Netflix’s data so you could drop it in your queue.
http://www.iheartmovies.com just added our Strands recommendation system and having some fun success. More on that soon!
Human powered recommendations can’t be beat.
A close match for Waltz with Bashir (2008) is Persepolis (2007)
Equally personal autobiographical memoir about life in Middle East also uses animation to tell its story.
http://www.cler...ir-2008/matches
Human Power can be biased by culture, taste, obscurity of movie, etc.
I think Netflix has a good automated program. If you want to take the average of all the Clerkdogs recommendations and give them to me, i’d be more happy. But trusting one human powered recommendation might not be best.
I just tried two common movies and it returned an ERROR for both. Do I have to include the date of the movie? If so, that’s lame.
Hi Matt,
What movies did you try? You do not have to include a date for the movie. We apologize for the error and appreciate your feedback.
Mary H
feedback@clerkdogs.com
http://www.clerkdogs.com
{seesmic_video:{”url_thumbnail”:{”value”:”http://t.seesmic.com/thumbnail/32XMc5WHPM_th1.jpg”}”title”:{”value”:” ”}”videoUri”:{”value”:”http://www.seesmic.com/video/wlN75E6CNl”}}}
Exactly. Maybe Netflix’s approach to invite people to come up with new collaborative filtering engine technology was a good approach a few years ago. Today, they could let developers come up with recommendation engines that people can use “in experience” with their Netflix accounts.
Thank you for the post. I know your website is about technology, but I appreciate the fact you take the time to add some fun. Keep up the great work!
Another movie recommendation site is http://www.jinni.com
It uses a bit different approach where you input your mood for the movie.
Jinni.com does generate movie recommendations, but they can be just as bewildering as Netflix’s. Type in “Slumdog Millionaire” and you get Charlie Chaplin’s “Gold Rush” and Robert Altman’s “Ready to Wear” (???). And they don’t even have “The Wrestler” in their system yet.
This is such a classic case of the Internet supporting not just a long tail, but a fat head that masks thoughtful human recommendations. I think Clerk Dogs sounds like a great service. Anytime you have a popularity algorithm, you will get pushed less quirky related choices. This is especially important for group projects like movies, where a single attached name (director, star, etc.) doesn’t carry the signal consistency in the same way that a single-author book does.
Mathew
http://www.blist.com
I highly doubt if human-curated recommendation system will work. It’ll be biased in my opinion. I guess the only solution for the best recommendations is crowd-sourcing, and websites like http://www.flixster.com for movies, http://www.cruxle.com for cross-media recommendations, http://www.ilike.com for music really prove this fact. My favorite is http://www.cruxle.com that recommends across media based on crowd-sourcing but in a very innovative way.
Even though clerkdogs might work for some movies, I highly suspect if it’ll work for all movies. One should wait and see.
Surf the Internet on your TV!
I’ve had the best (and most transparent) results for movie recommendations on http://www.criticker.com which basically matches people with similar movie tastes. Out of a pool of 20,000 users, it takes the ones that have a movie taste most similar to yours to base its recommendations on (out of a pool of 30,000 movies).
“Nanocrowd Movie Search: A New Recommendation Approach.”
http://primetim...search-new.html