To use Boolean Propagation:

NOTE1: Up to now, only binary variables are permited, one of the improvements I'll try to implement soon is the use of discrete (but not only binary) variables.

NOTE2: I changed computer recently and I'm having troubles with my Java compiler; thus, I couldn't upload an improvement I have that makes approximations for all the nodes. I will have it before december 20. Until then, this method approximates probability exclusively for terminal nodes (nodes without children). Sorry for the inconvenience.

1. Create a metaclass, and include the slot "Conditionals" of Float type in it (it must be able to contain multiple values)

2. The "graph" for the Bayesian network will be composed by classes. It is important that you pay attention in the order in which superclasses appear for a given class, for that order will be the base of conditional probability tables.

3. In each class, include a boolean slot having the same name of the class, with the suffix "v". Here is where you will insert the evidence. (You may also add all those slots on the terminal nodes, but I think it is easier that way).

4. On a terminal node (where you want to make inferences) include a string slot, and edit it to use BooleanPropagation (I wasn't able to implement a Float slot, for Java sent me lots of errors on all my tests).

5. In order to make inferences, you have to create an instance, and insert the evidence in the boolean slots. Then, press the button on the right, and the posterior probability will appear on the slot

BIG NOTE: The big improvement I want to do to the plugin is to create a Tab that makes it easier for the user to insert the Bayesian network and the conditional probabilities. Until then, you will have to do all the boring process described above :).