This work describes a novel approach to the problem of workforce distribution in dynamic multi-agent systems based on blackboard architectures, focusing especially on a real-world scenario: the multi-skill call centre. Traditionally, to address such highly-dynamic environments, diverse greedy heuristics have been applied to provide solutions in real-time. Basically, these heuristics perform a continuous re-planning on the system, taking into account its current state at all times. As decisions are greedily taken, the distribution of the workforce may be poor in the medium and/or long term. The usage of parallel memetic algorithms, which are more sophisticated than standard ad-hoc heuristics, can lead us towards much more accurate solutions. In order to effectively apply parallel memetic algorithms to such a dynamic environment, we introduce the concept of adaptive time window. Thus, the size of the time window depends upon the level of dynamism of the system at a given time. This research proposes a set of tools to automatically determine the dynamism of the system, as well as a novel and precise prediction module based on a neural network and a powerful optimization method.
Book Details: |
|
ISBN-13: |
978-3-639-71295-7 |
ISBN-10: |
3639712951 |
EAN: |
9783639712957 |
Book language: |
English |
By (author) : |
David Millán-Ruiz |
Number of pages: |
252 |
Published on: |
2014-04-17 |
Category: |
Informatics, IT |