Data-driven rank ordering — a preference-based comparison study

Data driven rank ordering refers to the rank ordering of new data items based on the ordering inherent in existing data items. This is a challenging problem, which has received increasing attention in recent years in the machine learning community. Its applications include product recommendation, information retrieval, financial portfolio construction, and robotics. It is common to construct ordering functions based on binary pairwise preferences. The level of dominance within pairs has been modelled in approaches based on statistical models, where strong assumptions about the distributions of the data are present. For learning pairwise preferences from the data we introduce a distribution-independent framework incorporating the level of dominance. We compare our approach with learning to rank order based on binary pairwise preferences through experiments using large margin classifiers.