nHow to formulate such a classification
scheme?
n
n1. How does such a classification relate to previous work?
n2. How are web queries
distributed given this scheme?
n3. How well do human
subjects classify against this scheme?
n
nJust briefly: how to automate such a classification?