Learning Binary Preference Relations:Analysis of Logic-based versus Statistical Approaches

It is a truth universally acknowledged that e-commerce platform users in search of an item that best suits their preferences may be offered a lot of choices. An item may be characterised by many attributes, which can complicate the process. Here the classic approach in decision support systems – to put weights on the importance of each attribute – is not always helpful as users may find it hard to formulate their priorities explicitly. Pairwise comparisons provide an easy way to elicitate the user’s preferences in the form of the simplest possible qualitative preferences, which can then be combined to rank the available alternatives. We focus on this type of preference elicitation and learn the individual preference by applying one statistical approach based on Support Vector Machines (SVM), and two logic-based approaches: Inductive Logic Programming (ILP) and Decision Trees. All approaches are compared on a dataset of car preferences collected from human participants. While in general, the statistical approach has proven its practical advantages, our experiment shows that the logic-based approaches offer a number of benefits over the one based on statistics.