Argumentation-Based Preference Modelling with Incomplete Information