Aggregation of Preference Data
By Yeawon Yoo
The session titled, “Aggregation of Preference Data” was held Tuesday afternoon – the last day of the INFORMS Annual Meeting. Even though the session was held at the last time slot, the room was full of many audience members and discussions. The session chair, Adolfo Escobedo from Arizona State University, opened the session by introducing the first presenter, Daniel Freund, who is currently a Lyft Research Fellow and will be an assistant professor at MIT in the Operations Management group next year. Daniel gave a presentation titled “Rank aggregation new bounds for MCx,” in which he introduced several Markov chain-based algorithms but argued that there had been no known performance guarantee. During his presentation, he proved the lower bounds on approximation algorithms.
Yeawon Yoo from Arizona State University then gave a presentation titled, “A new binary programming formulation and refined social choice property for expediting the solution to Kemeny ranking aggregation problem.” She introduced the new binary programming formulation and social choice-inspired decomposition that help to solve the ranking aggregation problem, which is known as NP-hard.
Adolfo Escobedo gave his own talk titled, “Joint aggregation of cardinal and ordinal evaluations.” He showed the axiomatic distances, new nonconvex exact formulations, and convex relaxations. With the real-world example of the 2007 MSOM Student Paper Competition, he highlighted the results of experiments when each judge submitted both cardinal and ordinal evaluations of their assigned papers.
The last presenter was Wenjie Tang, National University of Singapore. The presentation, “Elaborating or Aggregating? The joint effects of group decision-making structure and systematic errors on the value of group interactions,” introduced the laboratory experiments on checking if group interactions will have a positive effect on the accuracy of quantitative judgments. Depending on the existence of leaders, the aggregated outcome was different. Specifically, when there is a leader in the group, people spent more time discussing and had a higher motivation to systematically process information.