New Algorithm Can Find You, Even in Untagged Photos on Social Media Sites

It turns out that researchers may be able to find you on the Internet and learn about your relationships even if you aren't tagged in a photo. Scientists have created a new algorithm designed to tag locations to quantify relationships between individuals, even those not tagged in any given photo.

How does this new technology work? Imagine that you and your mother are pictured together in a photograph. You're both tagged in this particular photo and are quite close together. In the next photo, you're tagged with your father. Because of this close "tagging" relationship with both your mother in the first picture and your father in the second, the algorithm can determine that a relationship exists between your mother and your father and can quantify how strong it may be. In addition, if you are in a picture with your father and mother and only you and your mother are tagged, the algorithm can deduce the likelihood of your father being in that image.

"Two things are happening: we understand relationships, and we can search images better," said Parham Aarabi, one of the researchers, in a news release.

The new algorithm is called relational social image search. It achieves high reliability without using computationally intensive object or facial-recognition software.

"If you want to search a trillion photos, normally that takes at least a trillion operations. It's based on the number of photos you have," said Aarabi in a news release. "Facebook has almost half a trillion photos, but a billion users--it's almost a 500 order of magnitude difference. Our algorithm is imply based on the number of tags, not on the number of photos, which makes it more efficient to search that standard approaches."

Currently, the algorithm's interface is primarily for research. Yet it could be incorporated on the back-end of large image databases or social networks. This, in turn, could result in better results during researchers.

That's not all this algorithm is good for, either. The scientists also discovered that it can be used to generate maps. By tagging a few photographs of buildings around the University of Toronto and running them through the system with a bunch of untagged campus photos, they were able to create a pseudo-map of the campus from all of the photos they had taken.

The findings could eventually be included in social media sites in the future. For now, though, the new algorithm is just for research.