Naive Bayesian classifier have just two layers, one for Faults and the other for Symptoms. But, some researcher use Bayesian Network for classification such as the attachment.
On the other hand, based on Markov property, to find the probability of one node, it is enough to consider parents of the node (direct node connected to fault node). Therefore, in order to determine fault in inference stage, the adjacent layer of symptoms is needed. So, Naive Bayesian classifier should be work adequately. In this way, what is the roll of other symptoms nodes in the Bayesian Network?