Services Oriented Architecture provides Web Services (WSs) as reusable software components that can be applied to create more complicate composite services for users according to the specified QoS limitations. However, considering many WSs that may be appropriate for each task of a user-submitted workflow, finding the optimal WSs for a composite WS to maximize the overall QoS is an NP-hard problem. As a result, numerous composition schemes have been suggested in the literature to untangle this problem by using various metaheuristic algorithms. This paper presents a comprehensive survey and taxonomy of such QoS-oriented metaheuristic WS composition schemes provided in the literature. It investigates how metaheuristic algorithms are adapted for the WS composition problem and highlight their main features, advantages, and limitations. Also, in each category of the studied composition schemes, a comparison of their applied QoS factors, evaluated metrics, exploited simulators, and properties of the applied metaheuristic algorithms are explained. Finally, the concluding remarks and future research directions are summarized to help researchers in working in this area.

面向服务的体系结构将Web服务（WS）作为可重用的软件组件提供，可根据指定的QoS限制将其应用于为用户创建更复杂的复合服务。但是，考虑到许多可能适用于用户提交的工作流的每个任务的WS，为复合WS找到最佳WS以最大化整体QoS是一个NP难题。结果，在文献中已经提出了许多合成方案，以通过使用各种元启发式算法来解决该问题。本文对文献中提供的这种面向QoS的元启发式WS组合方案进行了全面的调查和分类。它研究了元启发式算法如何适应WS组成问题，并突出了它们的主要特征，优点和局限性。也，在所研究的组成方案的每个类别中，都对它们的应用QoS因子，评估指标，开发的模拟器以及所应用的元启发式算法的属性进行了比较。最后，总结了全文和未来的研究方向，以帮助研究人员在这一领域开展工作。