An Approach On Data Privacy Effecting Edge Modification In Social Network Graphs

Abstract

For the past few years, online social networks have been experiencing exponential newlinegrowth in the users and in the amount of available information. Many online social newlinenetworks offer web users, new interesting means to communicate and interact. In reality, information available on social networks commonly describes persons along with their personal information and interactions. Privacy preservation has an important role in publishing social network data online, without compromising any sensitive information. The information will not be misused by the general user but an adversary who wants to extract sensitive information and uses it against the user is a problem in the social era. Major research studies have been done for the generalization and suppression of sensitive information, such that identity is masked or unknown to others. However, there is always a compromise in the utility of the data. As the shortest path of the social networks holds the sensitive data, the shortest path or multiple paths are masked and the data is released to the third party. In this research work, the proposed edge swap method alters the edges of the shortest path between a set of targeted pairs given by the user. This is further extended to a proposed algorithm of K-shortest path for anonymity which creates n different similar paths of the shortest path with all edges in the path masked to provide higher levels of privacy preservation. The novelty of the third proposed Selective Perturbation method in this research work brings out the anonymization of the entire graph and maintains the original graph community based structures. The edge swap method basically swaps an non-visited edge with the partially-visited edge in the weighted undirected graph. Also the greedy perturbation supports in modifying the edges of the shortest path. The anonymization takes place by modifying edge weights on the shortest path with a condition to maintain the graph structure. The other edge weights of all-visited and non-visited edges are altered.