Description
Computational social choice theory is an emerging research area that studies the computational aspects of decision-making. It continues to be relevant in modern society because many people often work as a group and make decisions in a group setting. Among

Computational social choice theory is an emerging research area that studies the computational aspects of decision-making. It continues to be relevant in modern society because many people often work as a group and make decisions in a group setting. Among multiple research topics, rank aggregation is a central problem in computational social choice theory. Oftentimes, rankings may involve a large number of alternatives, contain ties, and/or be incomplete, all of which complicate the use of robust aggregation methods. To address these challenges, firstly, this work introduces a correlation coefficient that is designed to deal with a variety of ranking formats including those containing non-strict (i.e., with-ties) and incomplete (i.e., unknown) preferences. The new measure, which can be regarded as a generalization of the seminal Kendall tau correlation coefficient, is proven to satisfy a set of metric-like axioms and to be equivalent to a recently developed ranking distance function associated with Kemeny aggregation. Secondly, this work derives an exact binary programming formulation for the generalized Kemeny rank aggregation problem---whose ranking inputs may be complete and incomplete, with and without ties. It leverages the equivalence of minimizing the Kemeny-Snell distance and maximizing the Kendall-tau correlation, to compare the newly introduced binary programming formulation to a modified version of an existing integer programming formulation associated with the Kendall-tau distance. Thirdly, this work introduces a new social choice property for decomposing large-size problems into smaller subproblems, which allows solving the problem in a distributed fashion. The new property is adequate for handling complete rankings with ties. The property is leveraged to develop a structural decomposition algorithm, through which certain large instances of the NP-hard Kemeny rank aggregation problem can be solved exactly in a practical amount of time. Lastly, this work applies these rank aggregation mechanisms to novel contexts for extracting collective wisdom in crowdsourcing tasks. Through this crowdsourcing experiment, we assess the capability of aggregation frameworks to recover underlying ground truth and the usefulness of multimodal information in overcoming anchoring effects, which shows its ability to enhance the wisdom of crowds and its practicability to the real-world problem.
Reuse Permissions
  • Downloads
    PDF (2.6 MB)

    Details

    Title
    • Theoretical and Practical Advances in Computational Social Choice and Crowdsourcing
    Contributors
    Date Created
    2021
    Resource Type
  • Text
  • Collections this item is in
    Note
    • Partial requirement for: Ph.D., Arizona State University, 2021
    • Field of study: Industrial Engineering

    Machine-readable links