In the paper, I pointed out that it wasn't clear whether commonly used
predicate selection functions like entropy and the Gini index admit
efficiently computable pruning mechanisms. I'm glad to report that
this question has now been resolved in the positive: there is indeed
an efficient predicate pruning algorithm for all concave (convex)
functions, including entropy and the Gini index.

Please send me an email for the solution.

Kee Siong
12/12/05