Solving the Team Allocation Problem in Crowdsourcing via Group Multirole Assignment

In mass crowdsourcing, a platform is used for task allocation. Large complex tasks may be assigned to teams directly. To ensure the rapid accomplishment of tasks, we need to assign a task to multiple teams, while a team is composed of at least one worker. On the other hand, in order to ensure the enthusiasm and income of workers, the platform allows one worker to participate in a limited number of teams. Thereby, a team may undertake multiple but limited tasks. It is obvious that task allocation needs to avoid each worker being overloaded as well as prevent information leakage, which is caused by a task assigned to different teams at the same time. Therefore, different teams being assigned to a certain task cannot include the same workers. Such a scenario forms a many-to-many (M2M) assignment problem [or group multirole assignment (GMRA)] under high-order cardinality (HC) constraints, while the platform wants to choose appropriate and excellent teams for all the tasks in a specific time window. In order to solve this problem, this article studies and formalizes the above problem by introducing HC and conflicting agents on roles (CAR) constraints to GMRA. The main contributions of this article include: 1) the first formalization of the team allocation problem (TAP) in crowdsourcing through extending GMRA and the creation of a composition matrix to express HC constraints of agents and conflict avoiding constraints; 2) theoretical proofs of the theorems of the formalized problem, such as a necessary and sufficient condition (NSC), which confines the solution space of the problem; and 3) a practice solution to the proposed problem based on the IBM ILOG CPLEX optimization package (CPLEX). All the proposed approaches are verified by simulation experiments, which demonstrates that the proposed approaches are efficient, feasible, and practicable.