With the rapid development of Internet technology, online social networks (OSNs) has become one of the main ways for people to develop social activities. In order to maintain and strengthen interpersonal relationships, users are willing to share personal behaviors, feelings and other things through OSNs. Whether these resources reveal private information or not depends on the appropriateness of the access control policies set by the owner. However, with the increasing number of friends and complex relationships, it becomes more and more difficult for OSNs users to set appropriate access control policies. Aiming at above problems, a smart access control method for online social networks is proposed based on SVM algorithm to realize smart access control on the basis of integrating relationship types and description information of published content as eigenvectors. The experimental results show that this mechanism can automatically recommend the list of visible friends according to the content published by users and the relationship between users and friends, allowing users to modify the list to obtain the final access control policy, which can effectively protect users' privacy information. INDEX TERMS Online social networks, access control method, support vector machine, machine learning.

No abstract

No abstract

The rapid development of communication and network technologies including mobile networks and GPS presents new characteristics of OSNs. These new characteristics pose extra requirements on the access control schemes of OSNs, which cannot be satisfied by relationship-based access control currently. In this paper, we propose a hybrid access control model (HAC) which leverages attributes and relationships to control access to resources. A new policy specification language is developed to define policies considering the relationships and attributes of users. A path checking algorithm is proposed to figure out whether paths between two users can fit in with the hybrid policy. We develop a prototype system and demonstrate the feasibility of the proposed model.