A database framework for probabilistic preferences

Preferences are statements about the relative quality or desirability of items. Ever larger amounts of preference information are being collected and analyzed in a variety of domains, including recommendation systems [2, 16, 18], polling and election analysis [3, 6, 7, 15], and bioinformatics [1, 11, 19]. Preferences are often inferred from indirect input (e.g., a ranked list may be inferred from individual choices), and are therefore uncertain in nature. This motivates a rich body of work on uncertain preference models in the statistics literature [14]. More recently, the machine learning community has been developing methods for effective modeling and efficient inference over preferences, with the Mallows model [13] receiving particular attention [4, 5, 12, 17]. In this paper, we take the position that preference modeling and analysis should be accommodated within a general-purpose probabilistic database frame- work. Our framework is based on a deterministic concept that we proposed in a past vision paper [8]. In the present work we focus on handing uncertain preferences, and develop a representation of preferences within a probabilistic preference database, or PPD for short. This paper is an abbreviated version of our PODS 2017 paper, where an interested reader can find additional details about the formalism and proposed algorithmic solutions.