Rundensteiner, Elke A.
Ranking greatly assists us in decision making by allowing us to consider numerous options, with a variety of attributes and turn it into an understandable model. Yet, accurate rankings can be difficult for one to construct without extensive and proper knowledge on what you’re ranking. Thus arose the need for fair ranking systems that effectively elicit data from users and produce accurate rankings that satisfy and serve the users’ needs. In the pursuit of creating a fair ranking system this team proposes Rankit_Experimenter, a user study that tests the merits of three preference collection methods. Results indicate that categorical binning is the best value model: providing a substantial amount of data for the underlying ranking algorithm, while requiring minimal effort from the user.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.