Fp-tree Based Role Mining With Pre-assigned Weighted Permissions

The process of defining a complete, correct and efficient set of roles and associatedpermission is called Role Engineering. Role engineering has two approaches, top-downand bottom-up. The top-down approach is primarily business-driven, and roles are definedbased on the responsibilities of a given job function. The bottom-up approach is based onextracting roles from the existing access permissions using the data mining techniques.This bottom-up approach is called Role Mining. Most of the previous approaches arebased on Apriori data mining technique and treated all the permissions equally. None ofthe work has involved the pre-assigned weights and the frequentpattern growth (FP-growth) data mining technique based on frequent pattern tree (FP-tree). Thus providingmotivation for the current study.The current study is committed to role mining techniques to facilitate the security expertsand administrators to implement efficient Role Based Access Control (RBAC). Variousalgorithms are designed for role mining and most of them are based on apriori approach.In this study, we use a well-known data mining technique FP-growth based on the FP-Treedata structure, which is more fast and efficient than the other techniques used so far in rolemining. Apriori is slow as it uses generate pattern and test approach, i.e. first it generatesfrequent candidate permission set and then tests it that they are frequent or not. While theapproach based on FP-Tree role mining is fast both in space and time as it allowsdiscovering frequent item set without generating candidate item set.According to the needs of organization, we use the pre-assigned weights of permissions inthis approach. The weights are pre-defined by the concern organization in the light of theirsecurity model structure. Different organizations have different security levels and rankedthe permissions according to their own need. Our proposed approach added the pre-assigned weights to the role mining algorithm which is based on FP-tree data structure andFP-growth data mining technique. In the current research, we also proposed a novelapproach for calculation of permission-set weights, weighted support and weightedconfidence. This is a crucial enhancement in role mining field. In the current study, we propose an algorithm based on FP-tree with pre-assigned weightsof the permissions, which shows excellent results in terms of speed and efficiency. Theexperimental results of our work demonstrate the superiority of our algorithm.