Hi everyone, I am a Ph.D. student at UBC Vancouver. I am working with taxonomic hierarchies in probabilistic reasoning. We consider the domain of some of the variables is hierarchically structured (i.e., it can be represented in a tree hierarchy at different level of abstraction).
We are considering a very general BN where some of the variables are simple (have few values) and some of the variables are hierarchically structured. We can enter evidence at different level of abstraction and can answer the abstract query. I believe that it can be useful in diagnosis, decision making where we have to work at different level of abstraction. To test our approach we are looking some BN / real problems where some of the random variables have very large domain and can be represented in a tree hierachy form. Thanks in advance, Rita Sharma ------- End of Forwarded Message
