Compartilhar via


Two Posts from Viral worth a read

I don’t normally blog just links, but after Google took the chrome OS into the open last week I was going to put something up, I enjoyed the post from Fake Steve Jobs on the subject when it was first announced, but Viral has done a better job than I would have done.

Over the last couple of days I’ve been having a frustrating time putting videos together with a lot of time waiting for them o render: this would have been even more frustrating had Viral not sent me an invitation to Pivot.     He's written that up too.  I saw a post from Mary-Jo about Pivot a few days back and it is fascinating. It is a great way to slice through “proper” databases – anything that is a catalogue can be browsed any which way.

To avoid just throwing the links out there, let me just thrown in my tuppence on Pivot: it might get people thinking about Taxonomy. They’ll say “Wouldn’t it be great if we could that with Photos”; and then they’ll hit something of a wall. To visualize the problem I’d like you to think of St Paul’s Cathedral - which is on anyone’s list of the most iconic buildings in London. [A quick Bing search shows Paul is a popular enough Saint to have cathedrals named after him in quite a few other cities round the world]. Suppose you had a huge database with pictures of buildings. Which slices through that database might come up with St Paul’s ?  Buildings in London / Buildings in England (not Scotland or Wales) / Great Britain (not Northern Ireland) / The United Kingdom (England Scotland Wales, and Northern Ireland) / Europe. Places of Worship, Churches / Protestant churches / Cathedrals. / Cathedrals in capital cities. Buildings by Sir Christopher Wren. Buildings started in the 1600s. Buildings completed in the 1700s. Buildings built following the great fire of London. Buildings with domes. Classical Buildings. Important places in the life of [anyone buried / Married there]. Places mentioned in Films/Books. How do we get the information to make all the different slices we want associated with every picture ? I can see that if we geo-tagged it something clever could know the boundaries for London or Europe, and whilst we might put St Paul’s in a picture title, at a pinch the city or names of famous architects might make it to the keywords, but how many people would tag a picture of St Paul’s with “Nelson”.  It’s by no means clear how these problems can be solved, [I’ve got a personal taxonomy for classifying photos, but making it shareable is something else] however, that should not blind us to the data sets where problem has been solved already: they can give some fascinating results. I spent a little while with films and endangered species databases. You find yourself saying “I never knew …” which is the point of these things.