Preference Dominance Approaches for Conversational Recommender Systems

Conversational recommender systems typically involve iteratively showing the user a small set of options for them to choose between. In order to choose an appropriate set to display at each stage, it is useful to have information regarding the user’s relative preference between options: specifically whether an option is dominated by others with respect to the user’s preference relation. This paper describes a simple semantic framework for preference dominance, and two instances of the framework are developed for a recent kind of conversational recommender system. The first is based on a basic quantitative preferences formalism, where products are compared using sums of weights of features, and the input language involves comparisons between options. The second is a qualitative preference formalism, where models are a kind of generalised lexicographic order, and the inputs are comparative preference statements (in a language generalising CP-nets). A key feature of both methods is that deductions of preference dominance can be made efficiently, since this procedure needs to be applied for many pairs of products. The two approaches have been implemented, with encouraging experimental results.