Feature vector selection from a multi-dimension input object

Hi

I have a multi-dimension input object as follows

{ Id: 100, Date: '2013-06-06', Country: 'IT' attr: [ 'Andalus', 'Jokerman', . . (etc) ] . ., some other attr: ['Batang', 'xyz' ... (etc)] etc }

I am trying to think of mapping this kind of data structure to a feature vector which I can use to train some kind of a classification algorithm later.

What are some good strategies to obtain a feature vector from the above data set?

Thanks