Optimization of Automated Call Center Service Times Using Evolutionary Techniques

Call centers represent now an important channel for customer care activities. Most services provided by call centers are now accomplished in automated way to save operational costs. The customer is guided through a menu of options, which can be represented as a service tree, down to the desired service. In order to meet QoS constraints the service tree must be designed to minimize the average service time. An optimization algorithm, based on the combined use of Dandelion encoding and evolutionary operators, is proposed here to design the service trees with minimum average service time. The proposed algorithm provides trees whose average service time is lower than that obtained by the reference Huffman coding and is typically within 2% of the absolute lower bound represented by the entropic bound.