Foundations of preferences in database systems

Personalization of e-services poses new challenges to database technology. In particular, a powerful and flexible modeling technique is needed for complex preferences, which may even come from several parties with different intentions. Preference queries against a database have to be answered cooperatively by treating preferences as soft constraints, attempting a best possible match-making. We propose a strict partial order semantics for preferences, which closely matches people's intuition. A broad variety of natural preferences and of sophisticated preferences using ranked scores are covered by this model. Moreover, we show how to inductively construct complex preferences from base preferences by means of various preference constructors including Pareto accumulation. This preference model is the key to a new discipline called preference engineering and to a preference algebra. We present a collection of laws, including an intuitive non-discrimination theorem for Pareto preferences. Given the Best-Matches-Only query model we investigate how complex preference queries can be decomposed into simpler ones, preparing the ground for divide & conquer algorithms. We succeed to verify interesting adaptive filter effects of preference queries. Standard database query languages can be extended seamlessly by such preferences as exemplified by Preference SQL and Preference XPATH. In summary we believe that this preference model, featuring an algebraic foundation that matches intuition, is appropriate to extend database technology by preferences as soft constraints. Building efficient preference query optimizers, which can cope with the intrinsic non-monotonic nature of preference queries, investigations on how to e-negotiate in this preference model and a systematic approach to preference engineering are now feasible steps towards advanced database support for the ubiquitous real world phenomenon of preferences