PrefRank: Quantification and Aggregation of Subjective User Preferences

User-contributed content on the Internet has been growing at an extraordinary pace. Ranking vast amounts of such content, such as digital photographs, is handled well through user-driven ranking. It helps speeding up the ranking process while reflecting the opinions of the community. However, user-driven ranking can be often subjective and difficult to compare. We solve this using a well-known mathematical technique called the analytic hierarchy process. Due to the massive size of the user-contributed content, it is often not possible for all users to rank all items. Thus, finding a global ranking is a problem of rank aggregation of partially ranked lists. In this position paper, we propose a solution - PrefRank - based on eigenvector centrality that helps aggregating partially ranked lists. Our proposed approach can be used in other application scenarios involving qualitative judgement and ranking, such as reviewing academic papers for a conference.