The notion of an information tree has been formulated and investigated in 1982-83 by K.Chen and Z.Ras [1,2,3].In [1] we have defined a notion of an optimal tree(the number of edges is minimal) in a class of trees which are semantically equivalent and shown how optimal trees can be constructed.Rules can be discovered almost automatically from information trees.In this paper we propose a formalized language used to manipulate information trees,give its semantics and a complete and sound set of axioms.This complete and sound set of axioms is needed for discovering rules in information trees assuming that conditional attributes and the decision attribute are not arbitrary but they are both provided by the user.
展开▼
