in reply to XML probabilistic trees
Trees? Last time I studied AI we used probabilistic networks, AKA belief networks. That is, a node for each random variable, an edge for each non-indipendent pair of nodes, and a table of joined probability distribution for each edge.
Or, you may be talking about decision trees with stochastic discriminators...
In both cases, I'd use a standard algorithm on an in-memory representation, and then (when needed) serialize that representation as XML on filesystem or database.
-- dakkar - Mobilis in mobile
In Section
Seekers of Perl Wisdom