Flickr Photos Reveal Your Hometown?

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
According to a new study, researches can determine where you live by the photos you post on Flickr. What a stupid study, how can anyone tell I live in Las Vegas by this picture?

To prove their success rate, the study compares their guesses to the Flickr profiles in which users have published their home locations. “In 70 percent of the cases our algorithm has predicted the place of living of people with low error,” the study states.
Click to expand...
 
Is that statistic worded strangely? Low error in 70% of cases...and 60% of the time, it works every time.
 
I'm going to make a Flickr account and put up photos of the moon and Mars. Maybe even the ISS.

Let's see them figure that one out.
 
Kinda cool but pretty obvious. Next up on their list: they can tell where and when you take your vacation too....

Besides, I don't need flickr to tell me where someone's hometown is: I have facebook (stalking) for that ;)
 
This study is bull. I'm not the type of person who says "you should have paid me to tell you that", but in this case, yeah, you should've. So freaking obvious.

Steve said:
LOL, we should come up with a statistic for statistics that only work sometimes...
Click to expand...

Definitely perhaps. ;)
 
MrD1234 said:
It's more about shadows and geometry than anything else non obvious
Click to expand...

I guess if all your photos are taken indoors then it might confuse the algorithm. If you live 20 miles from the nearest small town and 50 miles from the nearest large city I wonder how well it works then.

I guess if you have accurate timestamps on the photos and they can use outdoor shadows then they could get fairly close on longitude and latitude.
 
GDstew4 said:
Is that statistic worded strangely? Low error in 70% of cases...and 60% of the time, it works every time.
Click to expand...


I thought that was a riot, actually...;) This: "In 70 percent of the cases our algorithm has predicted the place of living of people with low error,"...is written as obliquely and as strangely as it is because the author thought it better than writing: "In 45%-50% percent of the cases, our algorithm has predicted the place of living of people"....

Odds are it is something like that, with the algorithm slightly better than flipping a coin, assuming the results were pre-fudged to give each photo only two location possibilities...:D
 
That sounds like a very easy algorithm to design. Take the most common location that is tagged in your flicker account, and make that your guess. You could improve it by also looking at that data for that person's friends.
 
You must log in or register to reply here.
Back
Top