服务质量感知功能互补的可解释云API推荐方法研究

云时代，云API是服务交付、能力复制和数据输出的最佳载体。海量持续增长的云API严重影响了用户的选择决策，增加了云API的推广难度，从而使API经济的健康发展受到严峻挑战。本项目充分考虑云API推荐高质量、多元化和可理解的性能约束，从服务质量感知、功能互补和结果解释全新的视角开展个性化云API推荐方法研究。具体内容包括：(1)以获取的云API和服务质量数据为依据，挖掘用户与云API之间的交互关系，构建异构行为网络解决云API服务质量预测与推荐中数据表示与利用问题；(2)基于服务质量与情境信息之间的关联关系，设计特征选择算法提取有益情境特征，研究有益情境增强深度特征自动交互学习的云API服务质量预测方法；(3)在服务质量感知的基础上，面向多任务学习建立基于功能互补的云API推荐与文本解释生成模型。项目研究可为构建服务质量可感知、功能可互补和结果可解释的云API推荐系统提供理论依据和技术支持。

In the cloud era, cloud API is the best carrier for service delivery, capability replication and data output. However, the massive and continuous growth of cloud APIs has seriously affected users’ decision-making on cloud API selection, increased the difficulty of cloud API promotion, which challenges the healthy development of API economy. To cope with this problem, this project fully considers the high-quality, diversified and understandable performance constraints of cloud API recommendation, and carries out personalized cloud API recommendation method research from a new perspective of quality of service perception, functional complementation and result interpretation. The specific research contents of the project include: (1) based on the obtained cloud API and quality of service data, the interactive relationship between users and cloud APIs is explored, and the heterogeneous behavior network is constructed to solve the problem of data representation and utilization in cloud API quality of service prediction and recommendation research; (2) based on the association relationship between quality of service and situational information, a feature selection algorithm is designed to extract beneficial situational features, and then a cloud API quality of service prediction method based on deep feature automatic interactive learning enhanced by beneficial situational features is studied; (3) on the basis of cloud API quality of service awareness, a multi-tasking learning oriented functional complementation based cloud API recommendation and text interpretation generation model is established. The research of this project can provide theoretical basis and technical support for the construction of cloud API recommendation system with perceivable quality of service, complementary functions and interpretable results.