A classification-based approach for integrated service matching and composition in cloud manufacturing

Cloud manufacturing has been acknowledged as a transformative manufacturing paradigm aiming towards producing highly customized products via sharing distributed manufacturing resources and capabilities. One of the pivotal challenges regarding the practical realization of this idea is the process of matching manufacturing resources with personalized service demands. This problem contains two main steps: (1) retrieval of functionally similar services through assessing semantic similarity between resources' and subtasks' descriptions and (2) optimal composition of subtasks according to non-functional quality of service (QoS) indexes. However, almost all the research work in the field so far has focused on tackling each of these dimensions individually which barely represents actual conditions of the cloud manufacturing paradigm. To this end, this paper aims towards a novel integrated approach that first, successfully retrieves candidate sets for each corresponding subtask via implementing five classification algorithms and using TF-IDF (term frequency-inverse document frequency) vectors extracted from the manufacturing capability data. Then, optimal composite services are obtained for each scenario by using two well-known metaheuristic algorithms. Results obtained from the experiments have proven the advantages of this method resulting in a more comprehensive and realistic way for dealing with the service composition problems.