基于用户信誉度感知的Web服务在线个性化QoS预测方法研究

Web服务环境中由于存在恶意用户以及Web服务的更新变化，导致服务质量预测(QoS)准确性及效率难以提高，项目前期研究发现通过构建融合用户信誉度的QoS预测模型可显著提高预测准确性，但用户信誉度感知机制及Web服务QoS预测模型性能仍有待进一步探索。基于此，本项目以构建一种准确性高、自适应性强的Web服务QoS预测模型为目标，提出基于用户信誉度感知的Web服务在线个性化QoS预测方法，采用理论研究和实证研究相结合开展多方面研究，通过Web服务在线个性化QoS预测框架构建，用户信誉度在线感知建模及信誉度计算，基于用户信誉度及矩阵分解的Web服务在线个性化QoS预测建模等研究，将在结合上下文情境的用户信誉度多任务在线学习感知、基于在线学习的用户信誉度感知及矩阵分解在Web服务QoS预测中的应用形成创新。项目研究成果将可丰富Web服务推荐理论研究，并为构建高质量的Web服务系统提供理论参考。

In Web services environment, due to the malicious users and the frequent update of Web services, it may lead to lower accuracy and efficiency to the QoS prediction results. Previous research of project had found that the prediction accuracy could be improved dramatically by building the QoS prediction model which fuses user reputation with matrix factorization. But user reputation awareness mechanism and Web services QoS prediction model performance still need further explore. Base on this, this project aims to build a QoS prediction model with high accuracy and strong adaptability for Web services. This project proposes an online personalized quality of service prediction for Web services based on user reputation awareness. In this project, multiple steps of this approach are carried out by combining theoretical and empirical research. The contents of this project include: Web services online personalized QoS prediction framework, users online reputation awareness model and reputation computing algorithm, Web services online personalized QoS prediction model based on online reputation awareness and matrix factorization. The innovation will include user reputation awareness based on multitasking online learning in combination with context, the application of user reputation awareness fused matrix factorization based on online learning for web service QoS prediction. This project can enrich the Web service recommendation theory and provide theoretical reference for constructing high quality Web service system.