Does Unbbayes handle hybrid bayesian networks? if yes, should we use the EM algorithm to learn the parameters knowing that we have incomplete data or should we use another algorithm ?