A Clustering Algorithm Based on Rough Sets for the Recommendation Domain in Trust-Based Access Control

Abstract

Trust-based access control (TBAC) is a hot research issue in the area of network security for open networks. Clustering of the domains of recommendation entities is a prerequisite for trust quantification and evaluation in TBAC during the interaction of network entities. In this paper, we propose a clustering algorithm using the membership function based on rough sets for recommendation domain in TBAC in which traditional K-mode, k-mean and FCM clustering techniques are utilized. First, based on a system threshold for the number of recommendation entities, the proposed algorithm derives the clustering set for the recommendation entities within a most recent minimum time window. Then, the domain for the recommendation entity of the subject is obtained by analyzing the clustering results in which the entity data samples that the object gives to a request is fixed as the initial clustering centers in order to ensure the accuracy of the recommendation domain thus formed. To evaluate the effectiveness of the proposed algorithm, we apply both the IRIS standard data sets and the UCI data sets in our experiment. And our evaluation results show that the proposed algorithm performs better in terms of efficiency and clustering accuracy compared to traditional K-mode, K-mean and FCM clustering algorithms for recommendation domain in TBAC.