Using Causal Information and Local Measures to Learn Bayesian Networks