Based On Heuristic Analysis Of Large-scale Social Network Privacy Protection

Social network is often the abstraction of many real complicated network inreality, In the real world, the social network can be found everywhere, such as blogs,online social websites, and etc. The social network can provide people with somereliable, timely and high value information. At present, with the rapid development ofthe social network, the size of these networks enlarge rapidly, the data information ismore and more abundant, however these information often involve in user’s privacyand also people pay more and more attention to their privacy, so the privacypreserving measures need to taken for protecting user’s privacy information. Althoughsome traditional privacy preserving model have made some great progress, but thenetwork size is limited, rarely consider millions, ten millions, even billions magnitudescale of network, and also less give the attention to user’s personalized privacypreserving demand and the topology of network change largely, affect the available ofthe network data, and etc.This paper firstly introduces some basic knowledge about the social network andthe privacy preserving technology, then describes some classical privacy preservingmodels detailedly, these knowledges provide the theoretical direct for our proposedalgorithm.Based on above these describes, the fast privacy preserving algorithm oflarge-scale social network that we design based on the heuristic analysis mainlyincludes the heuristic analysis of the privacy preserving, the fast the privacypreserving，and etc. First of all, in view of the heuristic analysis of the privacypreserving, we design a new heuristic analysis function that is used for approximationthe optimal solution of the multi-objective function, reduce the change of the topologystructure; Then, design a fast privacy preserving algorithm for the large-scale socialnetwork, including the rapid community mining technology、 the personalizedk-degree anonymity、the generalization of the subgraph and the isomorphism of thenode. The algorithm divides the original network into a series of subnet (i.e., thecommunity structure) according to the fast community mining technique, and thencarry out the personalized k-degree anonymity for each subnet with less scalecompare with the whole network on distributed, then generalizes communities for super nodes and forms the condensed network. Then carry out the isomorphism of thenodes in condensed network, make the structure of all first-order neighbor nodes ofnodes with same degree to isomorphic. Finally, we carry out the simulation andexperimental on our proposed algorithm with some real social network, explain theperformance of the algorithm, and finally, realize the fast privacy preserving of thelarge-scale social network and user’ personalized privacy preserving demand, andreduce the change of the network topology structure.