There is a thought provoking post on the Go to Hellman blog about various metadata topics, library linked data and advances in facial recognition tools for identifying photo content.
Why is this important? The post raises some excellent points about how library data is created, how much data is optimal for optimization, and also why libraries need to think about promoting the findability of collections by employing search engine optimization techniques on existing library data.
As I have said before, any type of automated data – whether OCR text or Facebook photo tagging – requires an actual human to vet accuracy. Imagine the possibilities of image recognition software in an industry focused on visual resources.