Preference representation and reasoning is a key issue in many real-world scenarios where a personalized access to information is needed. Currently,there are many approaches allowing a system to assess preferences in a qualita-tive or quantitative way, and among the qualitative ones the most prominent areCP-nets. Their clear graphical structure unifies an easy representation of user de-sires with nice computational properties when computing the best outcome. Inthis paper, we show how to reason with CP-nets when the attributes modeling theknowledge domain have an ontological structure or, in other words, variable val-ues are DL formulas constrained relative to an underlying domain ontology. Wealso show how the computation of Pareto-optimal outcomes for an ontologicalCP-net can be reduced to the solution of constraint satisfaction problems
Reasoning with DL-based CP-nets / Di Noia, Tommaso; Lukasiewicz, Thomas; Simari, Gerardo. - ELETTRONICO. - 1014:(2013), pp. 640-651. (Intervento presentato al convegno 26th International Workshop on Description Logics (DL-13) tenutosi a Ulm, Germany nel July 23 - 26, 2013).
Reasoning with DL-based CP-nets
Tommaso Di Noia;
2013-01-01
Abstract
Preference representation and reasoning is a key issue in many real-world scenarios where a personalized access to information is needed. Currently,there are many approaches allowing a system to assess preferences in a qualita-tive or quantitative way, and among the qualitative ones the most prominent areCP-nets. Their clear graphical structure unifies an easy representation of user de-sires with nice computational properties when computing the best outcome. Inthis paper, we show how to reason with CP-nets when the attributes modeling theknowledge domain have an ontological structure or, in other words, variable val-ues are DL formulas constrained relative to an underlying domain ontology. Wealso show how the computation of Pareto-optimal outcomes for an ontologicalCP-net can be reduced to the solution of constraint satisfaction problemsI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
