MPI-I-93-260

In this thesis we are exploring some models for von Wright's preference logic. Given (initial) set of axioms and a set of formulae, some of them valid, some of them problematic (in the sense that it is not always intuitively clear should they be valid or not), we investigated some matrix semantics for those formulae including semantics in relevance logics (first degree entailment and RM3), various many--valued (Kleene's, {\L}ukasiewicz's, \dots) and/or paraconsistent logics, in Sugihara matrix, and one interpretation for preference relation using modal operators L and M. In each case, we also investigated dependence results between various formulae. Opposite problem (i.e.\ searching for a logic that satisfies given constraints) is also addressed. At the end, a {\tt LISP} program is presented that implements von Wright's logic as a decision supporting system, i.e.\ that decides for a given set of preferences, what alternatives (world--situation) should we choose, according to von Wright's preference logic system.