I've never heard before that there is a standarized format for representing Bayes Networks.
It's called BIF. It didn't fit my need 100% but I was able to import some basic information from BIF XML.
Some info:
http://www.cs.cmu.edu/~javabayes/Home/node7.html
Information I needed fom BIF was:
NETWORK[0].NAME - name of the graph
NETWORK[0].DEFINITION[0]
to
NETWORK[0].DEFINITION[7] --> nodes
NETWORK[0].DEFINITION[x].FOR - names of nodes
NETWORK[0].DEFINITION[x].GIVEN - edge list (backward) given in node names
NETWORK[0].DEFINITION[x].TABLE - CPT table for a single node - (Conditional Probability Table)
No comments:
Post a Comment
If you like this post, please leave a comment :)