基于进化算法的角色挖掘算法

发布时间：2018-04-03 19:38

本文选题：进化算法　切入点：基于角色的控制访问　出处：《北京交通大学》2014年硕士论文

【摘要】：基于角色的控制访问(RBAC, Role Based Access Control)是应用广泛的控制访问技术,较传统意义上的访问控制列表(ACL, Access Control List)具有很多优势,比如便于管理,消耗资源少等。从ACL模型转化到RBAC模型是一个富有挑战意义的过程,而解决ACL模型到RBAC模型转换的过程就是角色挖掘。本文首先分析了研究工作者目前提出的各种角色挖掘算法。其中,基于进化算法的角色挖掘产生了很好的效果,该类算法通过启发式搜索,寻找最小角色集合。然而这些算法在进行角色挖掘的过程中,没有考虑用户和权限本身所带的属性,这会造成实际应用出现问题,比如将医生的权限赋予计算机专业人员。本文从这一缺陷出发,提出两个新的角色挖掘算法,一个是基于遗传聚类的删除策略角色挖掘,首先根据用户和权限的属性产生候选角色集合,然后通过遗传算法来对角色进行删除,生成最终RBAC模型；另一个是基于蚂蚁聚类的添加策略角色挖掘,使用蚂蚁算法将符合属性策略的候选角色添加入RBAC模型。上述两种算法,在使用进化算法挖掘最小角色集合时,充分考虑用户和权限的属性,产生有意义的角色。实验结果表明新算法在充分考虑用户和权限属性的前提下,降低了角色产生的规模。
[Abstract]:RBAC (Role Based Access Control) is a widely used control access technology, which has many advantages over the traditional access control list (ACLs, Access Control list), such as easy to manage, less resource consumption and so on.Transforming from ACL model to RBAC model is a challenging process, and role mining is the process of transforming ACL model to RBAC model.In this paper, the role mining algorithms proposed by researchers are analyzed.Among them, role mining based on evolutionary algorithm has a good effect. This kind of algorithm finds the minimum role set by heuristic search.However, in the process of role mining, these algorithms do not consider the attributes of users and permissions themselves, which will lead to problems in practical applications, such as giving computer professionals the authority of doctors.In this paper, we propose two new role mining algorithms. One is deletion strategy role mining based on genetic clustering. Firstly, candidate role sets are generated according to the attributes of users and permissions.Then the role is deleted by genetic algorithm and the final RBAC model is generated. The other is the role mining of adding strategy based on ant clustering. Ant algorithm is used to add the candidate role according to attribute strategy into the RBAC model.The above two algorithms take full account of the attributes of users and permissions and produce meaningful roles when using evolutionary algorithms to mine the minimum set of roles.Experimental results show that the new algorithm reduces the size of role generation on the premise of fully considering the user and privilege attributes.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08

【参考文献】