Sunday, 17 March 2013

Let me tag that for you.

For years I've wanted to add some functionality I've had in mind, that really would make openDias truly useful. Automatically sensing that a new doc 'looks similar' to one you already have stored. Then, when it does, be able to apply the same tags and title to the new doc.

I've always had this down in the "would love to have" / "pipe dream", mostly because I couldn't think of a reasonable way of doing it. Well the other day during some idle brain time, some surfing got me to reading about media comparison algorithms for copywriter protection, and suddenly that long archived idea gets reclaimed as a "viable target".

Some focused Googling later and I find the pHash library. pHash is a perceptual algorithm that will provide a hash value for a media file. These hash values can be compared later to give a rating on how similar the two input files were. ........ Perfect.

A short hop and a skip later, we now have a 'would you like me to tag this for you sir?' option in openDias. I could not be happier.