Rough sets: the classical and extended views

The article is a comprehensive review of two major approaches to rough set theory: the classic rough setmodel introduced by Pawlak and the probabilistic approaches. The classic model is presented as a staging ground to the discussion of two varieties of the probabilistic approach, i.e. of the variable precision and Bayesian rough set models. Both of these models extend the classic model to deal with stochastic interactions while preserving the basicideas of the original rough set theory, such as set approximations, data dependencies, reducts etc. The probabilistic models are able to handle weaker data interactions than the classic model, thus extending the applicability of the rough set paradigm. The extended models are presented in considerable detail with some illustrative examples.