A relational framework for preference change

Abstract A rational agent typically adheres to a binary relation that represents their individual preferences, and thereby influences their choices and actions. This article introduces an applied formal framework for modelling the rational evolution (dynamics) of these preferences. Rather than committing to a specific preference logic (i.e. a formal object language and associated inferential machinery), the proposed framework is deliberately language-agnostic and adopts a relational (set-theoretic) approach. In this context, agent preferences are represented as arbitrary preorders over a set of items, with preference-change operations (revision and contraction) defined directly on these relations using set operations and closure conditions. The contribution is methodological and operational, as it provides explicit postulates and operator families on preorders that can be used directly when preference evidence is available as relational data (e.g. rankings or pairwise constraints). The study is complemented by the presentation of a graphical-based application, which implements the framework’s preference-change policies using the powerful Answer Set Programming paradigm.