Designing a consensus ranking algorithm for same domain entities

Conference groups, graduate admissions boards, and numerous social websites, deal in the commerce of grading content. The problem is that they have too much content for everyone to review. It consequently becomes essential to order content based on partial information. Merely relating numerical scores is erroneous because different assessors perceive different portions of the content and score in different ways. In this paper, we describe a Consensus Ranking Algorithm(CONSRANK) for the Linear Ordering problem explicitly intended for rank accumulation that can be used to construct a consensus ranking of choices. This ranking is based on reviewers' relative preferences of properties of options rather than of the options themselves. We then assess the performance of this Algorithm on data extracted from a number of academic graduate applications.