Core Concepts
Given a set of tasks and a set of voters with preferences over the order of task execution, the goal is to compute a consensus schedule that minimizes the average dissatisfaction of the voters.
Abstract
The content discusses the collective schedules problem, which involves scheduling a set of tasks shared by a set of voters (individuals) who have preferences over the order of task execution. Two models are considered: Order Preferences, where each voter provides a preferred order (permutation) of the tasks, and Interval Preferences, where each voter specifies a release date and deadline for each task.
The key highlights and insights are:
The problem can be solved optimally in polynomial time as an assignment problem, both for the binary criterion (whether a task is scheduled within the desired interval) and the distance criterion (how far a task is scheduled from the desired interval).
An analysis of the EMD (Earliest Median Date) rule shows that it is a 2-approximation for the total tardiness and total earliness criteria.
The paper studies the extent to which the rules (Distance Criterion, Binary Criterion, EMD) satisfy desirable properties like release date consistency, deadline consistency, and temporal unanimity when time constraints are inferred from the voters' preferences.
The paper also shows that the rules do not necessarily satisfy these properties, and provides examples to illustrate the cases where they fail to do so.
The problem is further studied under precedence constraints between tasks, showing that the previously studied rules can still be used with an additional polynomial-time step when the constraints are inferred from the voters' preferences, but become NP-hard when the constraints are not fulfilled by the voters' preferred schedules.