Abstract: How to accurately predict cloud service QoS and flexibly describe and handle QoS constraints are key problems of realizing reliable cloud service recommendation. For these two problems, this paper firstly provides a QoS prediction method based on case-based reasoning, and use this method to predict the QoS values of the candidate cloud services, so as to improve the reliability of cloud services recommendation; A QoS evaluation model based on Constraint Hierarchy is proposed, which divides the QoS constraints into hard constraints that must be met and soft constraining layers that allowed to be deviated according to the importance of QoS constraints for users, this model can effectively solve the over-constrained problem. The experimental results demonstrate that the proposed cloud services recommendation method is feasible and effective.

Key words: colud computing, cloud service recommendation, QoS dynamic prediction, constraint hierarchy

摘要： 如何精确地预测云服务的QoS，灵活地处理用户的QoS约束是实现可靠云服务推荐的关键问题。针对上述问题，提出了一种基于事例推理的云服务QoS动态预测方法，用于预测候选云服务的QoS值，提高云服务质量的可靠性；提出了一种基于约束层次（Constraint Hierarchy，CH）的QoS评价模型，该模型依据QoS约束对用户业务的重要程度，将其分为必须满足的硬约束层和允许被偏离的软约束层，该模型能够有效地解决过约束问题。通过实验验证了提出的云服务推荐方法的可行性和有效性。

关键词: 云计算, 云服务推荐, QoS动态预测, 约束层次