This paper introduces a formal base in order to talk about theory learning from a Description Logics (DL) perspective. A probabilistic Description Logics is introduced; and its need is intuitively justified. Theory learning is defined using information theory concepts and the probabilistic DL framework. Once this has been done, a general theory learning environment is constructed on this theory. This environment is based on the application of a set of induction rules. New rules can be defined easily using a set of syntactic manipulators that modify concept expressions.