Brian Smith, professor at Penn State, blogged about some work that Anurag, one of his graduate students is doing with scanning large media sets for “interestingness”:
Why I’m jazzed about this. I’m into the idea of using computing for experience capture, but the retrieval process will always be hard if mere mortals have to sift through every moment of their recorded lives to find interesting stuff. Anurag’s work is a step towards compressing and segmenting experiential data, making it accessible for those of us with little time to step through thousands of images to recall who we met on a particular day or where we left our keys. Regardless of experience capture, it just seems time to find ways to get machines to help us find interestingness in the daily onslaught of information that we get from email, Web sites, etc.
It does seem fascinating, to think about how to search for patterns in that much data — and also humbling and awe-inspiring to realize that our brains have all been doing that routinely our entire lives!
I must admit, though, that I felt an unusual (for me) luddite kind of counter-reaction. Not so much the conventional “such hubris to try to extract humanity into code” but rather “does my life really break down to such a small set of big chunks of daily routine that would be so easy for software to recognize?”
Yeah. It pretty much does.
In any case, recently I was kind of musing on the idea of running through Freebase with software, looking for patterns. I have to imagine the process is similar to what Anurag is doing, although algorithms for detecting interestingness in images and temperature data would be different from looking for patterns in the freeform database that is Freebase. I also have to imagine that Anurag has spent years learning about stuff I’m just scratching into. So there’s a counterbalance to the luddite thing.
Speaking of Freebase, I recently snagged an invitation and poked around, and there’s a lot of interesting stuff going on there. People are not clamoring for Freebase invitations like they have been for Joost, but I will probably have some things to say about it after I find some more time to explore. Alas, I don’t seem to have earned any invitations to pass along, but perhaps I’ll accrue some over time.