Go here and put your name in the box. Just do it. It’s awesome.
The project, called Personas, comes from the MIT Media Lab built by Aaron Zinman. Basically, it takes your name and searches the web for some context around it. It then takes the words and sites it finds to build a profile of your presence on the web. Or in MIT-speak using words like “corpus”:
Enter your name, and Personas scours the web for information and attempts to characterize the person – to fit them to a predetermined set of categories that an algorithmic process created from a massive corpus of data. The computational process is visualized with each stage of the analysis, finally resulting in the presentation of a seemingly authoritative personal profile.
Simply put, Personas represents the way the web sees you (or more specifically, your name). Of course, if someone has the same name as you, the results will vary, but for someone with a name like mine, it works very well. Basically, it says that “online” dominates my life, while I’m also associated with sports and books. Not really sure about the books part, but I am really into movies, perhaps it mistook some of my talk about a certain movie for me talking about the corresponding book (but there is a movies section as well).
But the process of it scouring the web for you name is probably actually cooler than the result. Watch it in action in the video below.
oh wow, this is impressive. I like the bottom thing:
HEADBUTT TO THE GROIN UNDERNEATH THE WATCH OF REFEREE MONET SAMUEL RYAN IS DOUBLED OVER AND RANDALLS COUNTERS WITH AN AXE KICK TO THE BACK OF RYAN’S HEAD
wow enter mg siegler that’s quite a negative impact you have on the web.. holy shit.. not even adolf hitler produces a result like yours..
It was sad watching my “fame” bar appear and then disappear :)
Sucks for the people unfortunate enough to have generic names– that’s why I’m naming my kid Xerxes Superfly
No suck for us common named folk. My name twins are CEO’s and published authors and such. I’m living vicariously.
I changed my name to Max Power for just this kind of stuff.
Stole if off a hair dryer
/homer
How about when Google thinks you’re actually “Rev Jeremiah Wright”? Seriously, half my results are political/religious/legal/illegal cause of this douchebag.
Le sigh.
lol…yeah, you win…I thought a real estate broker in New Mexico was obnoxious
Greg Brown FTL. Google = All sorts of Greg Brown, but none are me!
The big problem seems to be that it does what a Google search does and ignores the possessive, so my results are made useless by results like “Ian’s Crew” (e.g. “I’m so glad I’m not in class with anyone from Ian’s crew,” which is one of the ones that showed up as it was processing.) It seems like that would be something easy to fix.
BTW, Google has this problem even if you add quotes around my name. Sigh.
I feel your pain.
Their is a very popular international organization based on my last name (no relation, common English word)
“Increase your window size please. This requires 800px wide” – Get lost MIT? My laptop is as wide as anything! Ridiculous.
MIT using fixed pixels.
Priceless (Disappointing)!
Wow, that took a long time, but fun to watch.
I just did it. Page design is great and I found this an interesting site that I would share with friends. Would like to know more on how it comes up with this final bar graph though.
A new search engine and it could be promoted, though I think if you are associated with too much, it would just be a big mess. On the other hand, it could be a good tool to show your consistency and be judged on how well you stay on the bandwagon or jump onto others as other people do.
A drilldown on the results would make supplement my homepage very well.
It looks cool but comes up different every time I use it
It does vary each time, I wonder why…Too many variables in the formula maybe?
probably too much information on the web they can pull from.
So you are suggesting that it pull from different sites each time you do a query, or are you suggesting that search engines such as google is incapable of getting the same result twice in a row?
More like a bug to me — but still entertaining
You guys are right. “Paul Carr” comes up different on 4 different searches:
a**hole
jerk
loose cannon
narcissist
Man, Paul Carr is everywhere. lol
Yeah, mine comes up different each time too. although “sports” is consistently high. That’s because I share my name with a Samoan rugby player.
True, it varies everytime, its amazing to see how much of our personal info is already on the web to build such personas. Wonder how the results would be if they have access to the social networking profiles/activities !
@sriray
http://www.arktan.com
This is when I’m kinda disappointed in my parents for giving me the same name as the Texas revolutionary :(
ha ha yeah your name might not work so well…
You should try my name…
Am I missing something or can you not drill down at all? I saw something scroll by that I was credited in a game in 1993, a reference I did not recognize, and I’d love to find out what that’s all about.
btw, the creator is http://web.medi...t.edu/~azinman/ if you want to be more specific in the article :)
cool i’ll add his link in there.
I really wish I knew what the categories meant. Some seem obvious, but what is “illegal”?
yeah i caught that one too, but i noticed i was “legal” more than “illegal” which is good, i guess.
I seem to come up about even on the two. Wonder what that means?
If you search my full name which has very few results I can see it slowly analyzing, it sticks on website and analyses that a while and comes out with “Legal” and “Religious”. Seems it might associate specific words with certain things. For example if it finds “Warez” it might possibly count as Illegal. Things like that…
cool
This is a meaningless art project. Cool? I think not.
Sadly for me, there’s no digital trace of my existence. Google sees 443k instances of my name, and only 35.2k of my wife’s name, but personas loads that name just fine. (Helps that she shares a name with an artist).
Still very odd.
are you kidding me? did i break it?
whenever i put my own name in there, i get one entry that reads..
‘A DRUNKEN MICKEY SLATER IS MURDERED IN HIS HOTEL ROOM AFTER PICKING UP A VALUABLE BRIEFCASE’
i like to drink, but im not dead.. am i?
M!
Barak Obama is every persona.
not working on netbooks !! :-(
it says 7″ but even on my 9″ (Dell mini 9) I am just asked to expand my screen … and no way to use it
too bad
Even I end up getting different results every time and its very different each time.
SHALL WE PLAY A GAME? -Joshua
I liked it! Maybe it is not a brand new idea (ego search), but it is a different way of doing it.
Also, it was was well designed and the visual scheme is really good. Loved the colors. :)
And of course it takes a while to get the results… lots of info and crossed names around!
Pretty cool. But the results seem a little off, even for uncommon names. It seems to analyze various words and associate them with the categories. But words like talent, club, and career have fixed “sports” meanings despite their context having nothing to do with sports. Anyways, interesting though.
I ran this for Bill Gates. The results are bunk. What does Bill Gates have to do with Sports?
This is typical over-coverage of MIT’s work. This is typical, canonical praise for good presentation and mediocre results.
You’re totally right, the MIT brands it way too much for what its worth. Although people still are re-tweeting it like mad, I think only because a) its narcissistic in nature so it connects well with people, and b) people seem to find it mesmerizing.
However, you should note that the use of the underlying algorithm is intentional. Did you read that this is a critique of data mining? We knowingly used something that works better with less heterogeneity than the web gives us; it fit the design goal (not too many categories people could be in). It’s also my favorite technique of the moment :)
I wonder how they trained this. Did they run LDA on a bunch of sites and then assign names like media, fame, etc. to the resulting categories? Or did they create the categories and vocab weights first?
Great presentation, but you’re right, the results seem shoddy as does the premise of using people’s names for this.
Nope. Just generated lots of pairs of names weighted by the US census, then did the searches on them and built the model using LDA. There’s a lot of tweaks, such as removing Facebook & MySpace entries, etc.
Aaron
The idea is the main reason.
This thing sucks. I have “no digital trace” yet on google and yahoo all search results relate to me from page 1.
The queries are crafted in special ways to get snippets that are more ‘characteristic’ and meet various criteria.
It’s a nice light show, but useless if you don’t have a unique name (mine brings up me plus at least four other Len Feldmans.) Try “John Doe” and see what you get…
Turns out that most politicians – in the UK anyway – are pretty much all the same. Tony Blair manages some travel and design though:
http://www.inig...politicians.jpg
Doesnt work on my IBM T43/FF 3.5 which is a screen resolution of 1024×768 ( > 800×600).
Did you maximize the window? It won’t maximize itself for you. It really only needs 800×600 devoted to the flash concept.
I like the blurb in the youtube video that says ‘MG Siegler is just another militant Microsoft basher’. It made me lol.
bill clinton is a rapist apparently
I just tried Steve Jobs and though most things it was trawling through seemed to be for the correct Steve Jobs the end results were less than impressive. He is apparently big into family, fashion, social and education but not so much into design, media, management or professionalism. Back to the drawing board guys.
Great concept but needs some more stability. I tried it four times, with different results each time.. Also more details the results would be great, if presented in nice visual form..
I take a subset of the available information, which is randomized. It would be boring otherwise!
Next great Internet meme idea:
$ whois personaswar.com
No match for “PERSONASWAR.COM”.
It’s like kitten wars only more SEO geeky.
I was a captain. I was also convicted of treason and sentenced to death.
The illegal section on George Bush is bigger than his politics section…
Has anything of any significance ever come out of the work at MIT Media Lab ever during its entire history? Seriously.
Harmonix. And thus, Guitar Hero and Rock Band.
Letting me rock with the best of them alone has to count for something.
And e-ink, anti-aliasing, the original concept for Google Maps (done better), the electronic postcard, advances in codecs including JPEG & DSP, tangible interfaces, balhbalbhalbhalbh
Sounds like a generic probably jealous hater. Do you even realize this is an art project that’s backed by a stochastic model? Show me more new media that has technical sophistication while retaining aesthetic.
This is really cool looking. That’s about it.
Does not seem to be supporting non-Roman/Unicode characters? Tried in Japanese(日本語), Telugu(తెలుగు), Hebrew(עִבְרִית )..did not work. Aaron..if you read this plz enable multilingual support. But yeah..results got in English are interesting (searched for ‘famous’ nouns as currently my ‘digital traces’ do not exist!)
Sorry, but the model is built in English so that’s all I support.
Beautiful art, but a technology gimmick.
I convinced that the final graph has nothing to do with the actual search results. I suspect thant after the flash files loads with the results, it simply uses the text for some pretty animations.
I used some control names that are unique and have very narrow contexts and the final graph didn’t match the real persona or even the results the piece provided.
I cry foul! (But its fun anyway)
No gimmick, I visualize the process of doing inference using Latent Dirichlet Allocation. That’s what each color bar is, and you see it oscillate in assigning each word to a given category each iteration.
There was no promise the model was a good fit for your data whatsoever. The end result is the what the model infers from the text you see. If you read the description text, you’ll see that inaccuracies are part of the goal since this is a critique of data mining.
Personas ist a very enjoyable, colourful toy. Unfortunately it’s “results” are meaningless crap. I did a series of “searches” in a row, all to different results: see http://11k2.wor...du-im-internet/
well, I searched for my own name and got a bland “no digital traces found”. Google returns 1170 for my name in quotes, 13,300 without. There should be *something* out there in MITland?
I craft the queries in such a way that it tries to find ‘characterizing’ factoids about a person rather than just any web page that mentions the queried name. For example, if you enter my name you’ll see it has a lot less hits on Personas than a regular web search.
Aaron
Wow.. at first.. and maybe at last.
This site was fairly cool at the beginning.. I believe their introduction needs semicolon or two.. man the underratedness of the semicolon always amuses me. Anyways, it’s almost inspirational to type your name into the search bar and see where people who happen to share your name have succeeded in developing a web presence. Myself, in my many reincarnations, seem to be focused on sports and photography. While I am mostly a lettering high school athlete, I always meant to go further. While I enjoy photography, I never quite made it my passion. The passions I have? Refreshingly, they haven’t made there mark. More specifically, those who share my name haven’t taken to the spotlight yet in the fields I plan to. In a world where any kid can google something he’s into, find the best of the best at it, and potentially be discouraged by the disparity between his own skills and those of the world record holders; something like this can tell him that you know what? (s)He’s still something special in his or her own right. Deja vu right now btw….
It hates me. It returned only one result; accident.
but what is the point in it?
What a nice way to use an art/software project to brainwash superficial internet users. ;-)
Recipe: use those colorful MIT Media Lab stripes, make them animated as Latent Dirichlet Allocation model is being build to achieve results that are almost as misleading as the results of all those Facebook quiz apps. As I’ve just learned (kudos to author for that) LDA is a three-level Bayesian Model that evolves based on expectation-maximization algorithm. Wikipedia is everyone’s friend here.
Now, having in mind author’s comment on the Personas site and here, I wonder why this data mining criticism has not been emphasized a bit more. I’m sure MIT Media Lab students know more about social engineering on the internet and creating viral effects. :-)
The enter name screen has been updated with a mini-statement to do just that, and I’m working on a “read more” section to go into more detail. Problem is I’m probably at minute 14 out of my 15 minutes of fame. C’est la vie :)
Aaron
http://www.dailyperfect.com/ does internally almost exactly the same thing — it takes your full name and analyzes your digital footprint to create a profile of you. The profile then is applied to recommending content (the web site demonstrates by recommending personalized news). Here is a list of articles that describe the service in more details: http://www.asko...ws-site-launch/
Seems like they’re actually doing something different, at least at the low-level — from a quick usage it seems to splice up the words it finds into “concepts”, which then are used to create the newsfeed. Somewhat interesting approach, except that news recommendation systems have all kinds of philosophical problems. If they push their interface farther it could be a nice way to create a starter RSS reader, of sorts. Relying on names is highly problematic since so few people really have an Internet presence, let alone unique names. Personas attempted to highlight just that.
Yes, it won’t work excellent on all people, but there seem to be people reporting it works on them: http://twitter....?q=dailyperfect. For those with nonexistent footprint or too ambiguous names, tThe interface allows finetuning by thumbing up/down both for the topics/concepts, and individual news items, changing the news recommendation into more personal one in real-time. Seems
I saw two results which were actually about me: an email to an Apple developer mailing list, and the solution to an MIT homework assignment I submitted years ago (Aaron is obviously doing something other than a simple Google search).
Almost everything else treated my last name as a verb, as in “Hi Jamison, hope you have a nice birthday”.
I’m wondering if this is a similar result as http://qdos.com/?
Not at all. While I’m not going to make any claims that Personas is “useful” (that wasn’t the point), surely it is much more than qdos! George Bush has twice as much “individuality” as Barack Obama? How on Earth are they calculating these “statistics”, and what am I suppose to do with such information? At least Personas doesn’t pretend to be something it isn’t!