An MST-based information flow model for security in Online Social Networks

Abstract

Online Social Network (OSN) security issues have been thoroughly researched in the past decade due to their constant increase of users and ongoing feature development. The OSN is usually described as a graph, in which the users are the vertices, and their connecting edges represent relationships such as friendships, interactions and messaging. In our previous research we have devised an Information Flow security model that gives the vertices and edges numerical attributes, based on their user credibility and connection strength, such as age of user account and the friendship duration of two users, from which we have assessed the level of information sharing willingness of an ego node to users that are not directly connected to it, thus categorizing them as adversaries or acquaintances. In this research we use this model to generate a trustworthy network of users by creating a Minimum Spanning Tree (MST) of the graph instance and iterating the removal of the weak-attributed edges and their connected vertices. We use a known MST algorithm (Kruskal's algorithm) on this unique graph, thus achieving a good Trust threshold of a user network that allows a safe information flow in the OSN. Lastly, we validate the model and show its usefulness with experimental results.