A Naive Bayes Style

Naive Bayes classiiers can be seen as special probabilistic networks with a star-like structure. They can easily be induced from a dataset of sample cases. However, as most probabilistic approaches, they run into problems, if imprecise (i.e, set-valued) information in the data to learn from has to be taken into account. An approach to handle uncertain as well imprecise information, which recently gained some attention, are possibilistic networks. Because of the close structural resemblance of possibilistic networks to probabilistic ones, the idea suggests itself to construct a possibilistic classiier as a special possibilistic network in much the same way in which a naive Bayes classiier is a special probabilistic network. Thus we obtain a classiier that can easily handle imprecise information in the data to learn from.