混合云环境下面向多用户的云服务推荐方法研究

Over the last years, Cloud Computing becomes one of the most important service environments among the internet. In hybrid cloud, especially, the same cloud service can supply several different customers under the multi-tenancy mode, and the process of cloud service recommendation becomes more and more dynamic. . In this case, the service demands of various customers are hard to use in the recommendation process directly, and must be converted into resource demands. After organizing cloud services by various cloud resources, the optimal cloud service can then be chosen for different customers. Moreover, the diversity and correlation among different customers also have brought greatly challenges to cloud service recommendation. Therefore, multi-customers oriented cloud service recommendation methods in hybrid cloud environment should be studied.. In this proposal, some key issues in cloud service recommendation will be systematically studied, such as the diversity and correlation analysis among different customers, the models of organizing hybrid cloud service, and the recommendation methods for multi-customers. The detailed content is as following:. Firstly, the diversity and the correlation among different customers will be analyzed. On that basis, both the conflict measure model and the correlation assessment model will also be discussed.. Secondly, considering the diversity and correlation, the process of cloud service organizing will be studied. Service demands sometimes are easy to understand while sometimes are hard to read. On the basis of understanding of service demands, the discovery models for resource demands and its solving algorithms will be designed. And the organizing model for cloud service will also be established using resource demands.. Finally, the methods to recommend cloud service for multi-customers will be suggested. The schema will be changed automatically based on the complexity of the resource demands. And the recommendation model based on demand fusion and result fusion will be studied respectively. The dynamic negotiation recommendation model will also be set up in the face of the connection between different customers. These solving methods will provide a better cloud service to different customers simultaneously.. In summary, this project will not only expand the systematic analysis of hybrid cloud service recommendation, but also significantly advance the practically use for both cloud customers and cloud providers.

混合云环境下面向多用户的云服务推荐是一个十分复杂的动态过程。在混合云环境下，云服务提供商必须首先将以用户语言描述的服务需求转化为对各类云资源的需求，然后根据云资源需求组织出多个云服务方案，再优选出用户的满意方案，推荐给不同用户共同租用。由于不同用户的需求既存在差异又彼此关联，使面向多用户的云服务推荐具有更高的挑战性。为此，本项目从分析多用户间的差异性和关联性入手，研究多用户间冲突度量模型与关联程度评价模型；进而考虑多用户间的差异性和关联性利用多维数据研究基于用户服务需求的资源需求发现方法和基于资源需求的云服务的组织模型；在此基础上，分别研究基于资源需求融合的云服务推荐模型、基于方案融合的云服务推荐模型以及基于多用户协商的云服务推荐模型，并设计上述模型的求解方法。本项目的研究能够为解决混合云环境下面向多用户的云服务推荐问题提供新的方式和途径，对提高云服务质量具有重要的理论意义与实际应用价值。

本项目针对混合云环境下服务推荐过程的新特征，以云服务推荐方法为主线，从需求特征分析、服务资源组织、服务动态推荐以及典型领域应用等方面开展了研究工作。. 围绕混合云环境下多用户的交互过程，分析了多用户需求的差异性和关联性，建立了基于证据理论的多用户需求融合模型，提出了多用户需求冲突检测模型与冲突度量方法，提出了面向领域的多用户需求偏好建模方法。同时，考虑混合云环境下服务资源的组织过程，设计了一种包括基础架构层、资源管理层、资源需求调度层、用户界面层以及应用层的云服务架构，分析了虚拟服务资源的组织与分配过程，提出了基于负载均衡策略的云服务资源组织模型及优化求解算法，建立了考虑时间成本、费用成本和能耗成本的云服务资源分配模型及其优化求解算法。在此基础上，设计了集成协同过滤推荐算法的混合云服务动态推荐系统，该系统包括用户模型库、离线推荐列表、动态推荐列表三个核心功能，并行采用多种推荐算法生成云服务推荐列表，实现多用户需求的动态融合与云服务的动态更新。. 在开展理论研究的过程中，项目组还关注到新一代信息技术对混合云计算领域带来的重要影响。结合混合云在典型领域应用过程中蕴含的管理科学问题，围绕有线电视数据云服务、警务智能数据云服务、无人机管理云服务等领域中有关服务推荐的关键科学问题，进一步针对混合云服务动态推荐方法与技术开展了理论研究，为上述典型领域的混合云服务应用实践提供了一定的理论和技术支撑。. 在研究过程中，项目组成员在国内外重要学术期刊或会议发表学术论文20篇，其中被SCI、EI同时收录10篇、EI收录4篇，在基金委管理学部认定的中文期刊或CSSCI收录源刊发表学术论文4篇；参加学术会议13次，进行分组报告13人次；资助3博士研究生和4硕士研究生开展相关研究工作；获授权国家发明专利8项，计算机软件著作权1项；安徽省科技进步奖二等奖1项。