Automated Optimal Dispatching of Service Requests

In the services domain, the customers raise issues and service requests in the form of tickets. There is a pool of personnel who work on these tickets and resolve them. The problem at hand is to dispatch these tickets to the most appropriate personnel. Optimality is applied to metrics like the mean service time taken to resolve a ticket, the fair sharing of workload among the personnel, and the size and configuration of the pool. The current state of the art involves a human dispatcher for assigning incoming service requests. Though intelligent, a human dispatcher can be suboptimal with respect to the above mentioned objectives due to the large space of parameter values to be considered. Further, there exists an opportunity to achieve high-level goals such as on-the-job training, eliminating overproduction, and workload balancing among personnel through smarter dispatch decisions. For example, target skill levels of personnel can be achieved by assigning them tickets requiring those skills increasingly. Also, overproduction can be controlled by dispatching only those tickets that otherwise would be in the danger of missing deadline (SLO) constraints. Our work involves the design and implementation of an automated dispatcher which would take various characteristics of the tickets and the pool state as input and recommend an intelligent dispatching decision for the tickets, based on the above-mentioned goals and constraints.