Contextual ranking of query results with incomplete preferences

Context-aware database is frequently used in user-centric applications. The users normally tend to express their preferences through comparisons. For example, Bob prefers Cornell to USC. Because the users only compare a small proportion of the products under difference contexts, ¿incomplete¿ preferences are a common occurrence. We propose a ranking approach which can contextually rank the query results when the users preferences are incomplete. We empirically evaluate the practical effectiveness of our method.