Rate my startup - deegr
With a couple of Uni friends, we have been working on http://www.deegr.com for the past few months; it's a cultural recommendation engine that lets you find people with similar tastes on the whole Web.
From the landing page, just rate 5 movies you either like or hate. deegr instantly searches through over 60k+ user profiles we've referenced on websites such as IMDb, Rotten Tomatoes, or AlloCine (and deegr itself!), to show you the 30 people that best match your tastes and give you inspiration based on the rest of what they've also liked. The more you rate, the better the accuracy of your search!
You can also follow your own Facebook/Twitter crowd, share micro-reviews, interact with other deegr members and a few other things after a free, flash-registration.
The purpose of this project is to create a generic recommendation engine for pretty much anything that has ratings (or the equivalent) on the Web.
We'd love to have your thoughts and suggestions on it: the idea, the UI, the quality of the recommendations, the speed of the engine... Anything.
If you also have suggestions regarding blogs & sites we should get in touch with so they can spread the word about deegr, don't hesitate!
Thanks all,
The deegr Team http://www.deegr.com
ps: other medias (music, books, videogames...), the API and mobile apps are on the roadmap; just give us a few months!
EDIT: Please note that we're currently working on a small feature that will enable you to minimize mainstream movies in a single click, in our recommendations! We know you've most probably seen Shawshank Redemption and the Godfather ;)
50 comments
[ 2.4 ms ] story [ 111 ms ] threadWhat happens if you ask people to rate 10 movies with either a thumbs up or thumbs down? Would that give better results?
It's quite interesting, although I could see more use for it as part of other websites. EG: dating websites could use the tech to match people?
Good Luck though!
Well, we will see how things evolve. You are correct in that it seems like people are more likely to give 5/5's than lower ratings; but I guess that's just one way of matching people: they can be similar in their way of rating things (they could love everything or be big time haters), deegr will match them. Also, it's not like we can't interpret ratings over 3/5 as thumbs-ups and lower ratings as thumbs-down; we can still change, only time will tell us what the best scale is. As of now, the current system seems to do a great job.
And yes, of course, we're thinking a lot about implementing it on other websites; we're already working on an API ;)
Thanks for the feedback!
* I LOVE that you don't require a sign-up to get started. The fact that I can start rating right off the front page is equally good.
* I don't grok how the right side "Recommendation Summary" widget works. I rate stuff and they... disappear? Wait, aren't these supposed to be recommendations in the first place? So where are the ratings (starts)? Also, what's the relation between this widget and the content to the left (my "circle")?
* To the left, I'm seeing some dude's ratings. Maybe you should change the graphics for dude's ratings and my ratings? Because both are in blue.
* As I mention I already use several recommendation engines. I would really love to submit my Criticker profile page and have the ratings imported. Because you know I have spent hours rating stuff over the course of several years.
* I don't see the purpose of "You both rated : (20)" In the left content pane. Also, it takes up a lot of visual space.
* This is pretty fun to use.
* Some of the UI - sub-items of the Movies tab; the hover for the Register link - are still in French.
* I've seen most of the recs I'm getting. Again, if there was a way to import ratings from other sites it would make a huge difference.
* 2 and * 3: the left side shows your different matchs' ratings (what they've liked, what they haven't). The right side shows a reorganized summary of all the movies liked by the 30 people whose tastes best match yours. It's all explained when you hover your cursor over the top of that column ;)
* Yes, we've thought about that. I can't tell yet if we're going to work on that soon or not; it would be pretty easy since we've crawled the profiles already, but that's kind of unfair to the other websites, it feels like we'd be stealing profiles - whereas right now, we only reference them.
* 'You both rated' helps you have a quick glance at how well you and the person in question have in common. We've tried without it, and quite frankly, it's better with; juste to know on what kind of movies you both match. The number in parenthesis is the total number of movies in common with similar ratings.
* ^^
* Woopsie! We'll fix that asap!
* Yeah, it's not the first time we're told the Recommendation Summary shows some mainstream movies. Well that's because they've liked a lot, so pragmatically, we HAVE to recommend them; and if you liked/hated them, rate them away!
HOWEVER please note, we're only weeks away from an update that will allow you to hide Mainstream recommendation in a single click, so please be patient!
Thanks again for that full review! And don't forget: the more you rate, the better the accuracy.
I actually found a new movie that related to the ratings I supplied, nor have I seen it before, good stuff.
Here are some thoughts:
1. I've feed Netflix a lot of my data over the years. A link from each movie to a Netflix result would show how Netflix rates it related to me. Maybe you don't want to push people to Netflix, but it's useful information to me.
2. I don't really care about the "people" I care about the movies. As I click through the results I see a lot of duplicate movies under different people. Remove or leave the people category, I don't care, but strip away the duplicate results...
All in all I find it useful.
1. Oh yes we want to push Netflix! We're just a small french project and Netflix has not invaded us as of this date; and is unlikely to do so for sometime, actually. I know what you're talking about and we'll see how/if we can implement it.
2. Well that's a bummer, deegr's all about finding others like you ^^ If we only recommended movies without telling you why you should like them, we'd ressemble every other engine out there! Don't you like knowing you have taste-mates out there?
Cool stuff though. I really like where your going with it. I instantly found it useful to find a new movie which is what is the most important to me now ( tomorrow following taste mates may trump that )
At one point, the text at the top started saying "Rate NaN movies".
You seem to start off with extremely popular movies. Those movies seem unlikely to generate a range of ratings. You might consider starting off with movies that a large number of people have rated but which polarize people, which would help differentiate people sooner and improve the quality of the circles.
At one point the "I didn't see this movie" links became french: "Je n'ai pas vu ce film".
In one category, I hit "I didn't see this movie", and later the same movie showed up in another category.
I found it slightly odd that the rating scale goes from 0.5 to 5 stars, rather than 0 to 5 stars.
I rated all the films deegr showed in the first batch, and they all disappeared, but now I don't see an obvious next step. The rotating deegr explanation banners at the top mention something about "refine your search after rating a few movies", and shows an icon for that, but I don't see that icon anywhere in the interface. I finally found that clicking on the "Deegr Circle" dropdown, then "Deegr Circle" under that, led me to something useful.
Once I get to the mode showing a circle of other people, I don't see any options for "I haven't seen this film" anymore.
Once you start recommending a circle of people, and recommending other films liked in that circle, that encourages rating those films too, which seems unlikely to change the circle.
I'd love to see an explanation of the name. Best guess: modified form of "dig"?
We'll take some time this week to search & fix the translations you reported.
As for the mainstream recommendation, I've edited the original post: we'll soon let you users get rid of mainstream recommendations in a single click.
Regarding the scale starting at 0 instead of 0.5, that's because you have to give them a .5, that's for several reasons: for example, you wouldn't know if you've rated the movie a 0 or not when clicking somewhere left of the stars. Also we think it's more disgraceful to get a 0.5 than a 0. And other reasons.
Regarding "but now I don't see an obvious next step.", this sounds weird; you should normally have been redirected to a search result (a Circle) straight out of rating 5 movies, no more. Isn't that what happened?
There's no other "I haven't seen this movie" once you land your first search Circle, because there's no point. Either you haven't seen it and it stays as a recommendation (which you can add to your deegList), or you really don't want to see it and you can therefore Mute it with the "-" sign.
Yep, deegr, derived from digging. It's quick, catchy, simple, easy to type and to say... And it was available :/
As for 0.5 versus 0, fair enough; it just seemed a bit quirky. :)
> Regarding "but now I don't see an obvious next step.", this sounds weird; you should normally have been redirected to a search result (a Circle) straight out of rating 5 movies, no more. Isn't that what happened?
Nope; I kept rating movies, until eventually all of the initial sets of films disappeared, and nothing else happened.
> There's no other "I haven't seen this movie" once you land your first search Circle, because there's no point. Either you haven't seen it and it stays as a recommendation (which you can add to your deegList), or you really don't want to see it and you can therefore Mute it with the "-" sign.
Fair enough. However, I wonder: do you take "mute" into account when doing recommendations, or just hide films? In particular, I see a number of films for which I could either rate them 0.5 ("I can tell from the trailer that this will suck") or mute them ("I guess I should reserve judgement, but I don't ever want to see it").
One more misplaced translation: "Retour en haut".
No, we don't currently take into consideration movies that have been muted or added to the deegList, but just like the polarization, we've already given thoughts about it (especially for this) and will probably take it into account in the next algorithm update. Thanks again ;)
Note the second and third cases under "middle ranking". That seems like a category of films you should present to help add more nuance to a user's ratings and circle.
It takes literally no more than 10 seconds to create an account, you should give it a go ;)
After I started rating movies that I've seen, I could see the value in the recommendation engine aspect...
I'm not a fan of the name either... it's not memorable.
Also, it seems that French text shows up every now and then.
Good start, keep iterating.
Problem solved is giving people a reason to actually trust the recommendations made by recommendation engines. We have never known why Amazon (IMDb etc) was recommending us some movies rather than some others. There's a very strong algorithm behind these recommendations, but we believe that humanized recommendation engine is better because you can actually ask the person why he/she liked a movie and would recommend it. It's kind of like Aardvark: sometimes you want a person, not a web page, to answer your question.
How about that? At first it shows-up like words; once you've clicked one, it only ever comes back as an icon. ?
Also, clicking on it once doesn't mean a user will know or remember what it does. :)
Looking into fixing this asap. Thanks
Btw, deegr is muuuuuuch smoother on Chrome or Safari.
Secondary comments/suggestions:
1. After I voted a few movies I could see a list of similar people and a number of suggested movies. I could also see the movies other people liked, their votes and vote for them. As an UI suggestion make sure that when I put my mouse over another user's vote I understand that I can vote -- currently is not obvious
2.There is some french here and there...
3.I would love to see some simple "Watch this..." screen that is centered more to suggested movies and less on similar people. After all I am here to find movies!
Maybe you will actually find it useful to find people like you when you can find people like you from your town/age/sex/opposite sex... See what I mean? ^^
I would only suggest that you give people the option to rank more movies before building the circle if they want to. I think I could have gotten even better results if I could've reviewed more films.
Hope you succeed!
http://www.betabeat.com/2011/10/26/quit-netflix-meet-the-net...
I've just now been using it, and it's pretty interesting...