Intelligent and Mined-knowledge Driven Consultant Services for User Behavior Recovery in Large Distributed Systems

大型分布式系统中用户行为恢复的智能和挖掘知识驱动的咨询服务

• 批准号: RGPIN-2014-06402
• 负责人: Sartipi, Kamran
• 金额: $ 1.46万
• 依托单位: McMaster University
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2016
• 资助国家: 加拿大
• 起止时间: 2016-01-01 至 2017-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=612199
• 关键词: Intelligent; Mined; knowledge; Driven; Consultant

An increasing number of organizations are integrating their public services to become more competitive in providing broader and more easily accessible services to the public. The resulting federated organization must relax its internal regulations and access control policies to provide compatibility of services that collectively form a trust model of service usage. Such trust models jeopardize the public’s precious data assets. The complex and changing nature of usage patterns and the high costs of violating a trust model by accessing sensitive private data, create serious threats to the integrity of distributed information systems that are the backbone of public services. The proposed research program aims at providing a new generation of intelligent decision support services, namely “Consultant-as-a-Service” that effectively assists and guides a broad range of users that can be categorized as: i) customers of organizations who are not familiar with the large variety of organization’s services to the public; and ii) system administrators of large distributed systems who are overwhelmed by the number of system users whose behavior in using the system services must be monitored and controlled. Most large organizations are operated by outdated legacy systems that are not compatible with advanced technology for secure communication, and protocols for user authentication and authorization. Therefore, the system vulnerability problem is becoming more serious and may drastically impair the quality of service provisioning. This issue may cause vast economical costs for the organizations whose private data assets have been compromised. For example, the current state of distributed medical imaging systems and their central repositories in most provinces is based on a trust model, which makes it vulnerable. Such cases have been the main motivation for the approach in this proposal. The proposed consultant service is knowledge driven and closely assists the security administrators to compose detailed and complex patterns of usage and provides solutions that can answer to a variety of user behavior questions, which will save valuable time and effort that must be invested for such discoveries (if possible at all). Example queries are as follow. What is common among users who accessed the system around the traffic rush hour? Which policy rules are sensitive to the current state of system usage? What is the frequent behavior pattern of a specific role or user in the system? Is there any behavior similarity among users in specific and separated locations, and in what aspect they are similar? Etc. Furthermore, there are huge possibilities to identify anomaly behavior of some users by close investigation of the identified common behavioral patterns, in which the existence of some actions cannot be justified by the related policy rules. The proposed research will train three Master’s students who will become expert in different skills such as: applying data mining techniques on different data assets; working on advanced decision support techniques; challenging with high complexity of optimization search algorithms and constrain pattern matching; and designing pattern languages. Two PhD students will be trained in this research, which will provide theoretical and practical solutions to novel user behavioral modeling and designing intelligent decision services. All research students will have access to an advanced cloud based data centre that provides an experimental distributed medical imaging system to design, develop and evaluate their research works.

越来越多的组织正在整合其公共服务，以便在向公众提供更广泛和更容易获得的服务方面更具竞争力。由此产生的联邦组织必须放松其内部规定和访问控制策略，以提供共同形成服务使用的信任模型的服务的兼容性。这种信任模式危及公众宝贵的数据资产。使用模式的复杂性和不断变化的性质，以及通过访问敏感的私人数据而违反信任模型的高昂成本，对作为公共服务支柱的分布式信息系统的完整性造成了严重威胁。 建议的研究计划旨在提供新一代的智能决策支援服务，即“谘询服务”，有效地协助和指导广泛的用户，这些用户可分为：i）不熟悉机构向公众提供的各种服务的机构客户;以及ii）大型分布式系统的系统管理员，他们被系统用户的数量所淹没，这些系统用户在使用系统服务时的行为必须被监视和控制。大多数大型组织都由过时的遗留系统运营，这些系统与安全通信的先进技术以及用户身份验证和授权协议不兼容。因此，系统脆弱性问题变得越来越严重，并可能严重损害服务提供的质量。这个问题可能会给私有数据资产受损的组织带来巨大的经济成本。例如，大多数省份的分布式医疗成像系统及其中央存储库的当前状态是基于信任模型的，这使得它很容易受到攻击。这种情况是本提案中所述方法的主要动机。 建议的顾问服务是知识驱动的，并密切协助安全管理员组成详细和复杂的使用模式，并提供可以回答各种用户行为问题的解决方案，这将节省宝贵的时间和精力，必须投资于这样的发现（如果可能的话）。示例查询如下。在交通高峰期访问系统的用户之间有什么共同点？哪些策略规则对系统使用的当前状态敏感？系统中特定角色或用户的常见行为模式是什么？在特定的和分离的位置上的用户之间是否存在行为相似性，以及它们在哪些方面相似？此外，存在通过对所识别的共同行为模式的密切调查来识别某些用户的异常行为的巨大可能性，其中，某些动作的存在不能被相关的策略规则所证明。拟议的研究将培养三名硕士生，他们将成为不同技能的专家，例如：将数据挖掘技术应用于不同的数据资产;研究高级决策支持技术;挑战高复杂性的优化搜索算法和约束模式匹配;设计模式语言。两名博士生将在这项研究中接受培训，这将为新的用户行为建模和设计智能决策服务提供理论和实践解决方案。所有研究生都将可以访问先进的基于云的数据中心，该中心提供实验性分布式医学成像系统来设计、开发和评估他们的研究工作。