SC-MA: A Service Composition Method Based on Multiple Adaptive Mechanisms

Web service combination is the integration of numerous existing independent services using specific logic and processes to meet more sophisticated business requirements and functions. This combination can be easily customized to various application scenarios to achieve business automation and efficiency by coordinating the inputs and outputs of individual services. Adaptive mechanisms play an important role in service combination. In this paper, we introduce a service combination framework called SC-MA which uses several adaptive mechanisms to tackle the service combination problem. We divide the flow of the algorithm into three main steps to achieve efficient and accurate task solution construction. First, we analyze the relationships between services and estimate the likelihood of each service's occurrence using a graph neural network (GNN). After obtaining high-probability predictions of services, we combine Pointer Networks (PNs) with adaptive mechanisms to construct preliminary solutions. Finally, we introduce the Whale Optimization Algorithm for Adaptive Mechanisms (CXWOA) to refine the initial solution, aiming to achieve a higher-quality result. The experimental results suggest that the adaptive mechanism can improve service portfolio performance by enhancing average service quality while decreasing average runtime.