International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169 Volume: 2 Issue: 11 3647 - 3649 _______________________________________________________________________________________________ 3647 IJRITCC |November 2014, Available @ http://www.ijritcc.org _______________________________________________________________________________________ Review Paper-Social networking with protecting sensitive labels in data Anonymization Miss.P.S.Kadam #1 , Prof.S.V.Patil #2 , Miss.A.A.Bhosale #3 , Mr.D.H.Dewarde #4 #1 Computer Sci. and Engg.Department,MalajirajeBhosale Technical Campus, Islampur. #2 Computer Sci. and Engg.Department, AnnasahebDange College of Engineering, Ashta, Sangli. #3 Computer Sci. and Engg.Department,JaywantCollege of Engineering&Management,Killemachindra Gad #4 Computer Sci. and Engg.Department,MalajirajeBhosale Technical Campus, Islampur. E-mail ID: #1 pallavikadam14@gmail.com, # 2 siddheshwar.patil@gmail.com, #3 bhosale.anita11@gmail.com, #4 dewarde.digambar73@gmail.com #1#2#3#4 Shivaji University, Kolhapur, Maharashtra, India. Abstract--The use of social network sites goes on increasing such as facebook, twitter, linkedin, live journal social network and wiki vote network. By using this, users find that they can obtain more and more useful information such as the user performance, private growth, dispersal of disease etc. It is also important that users private information should not get disclose. Thus, Now a days it is important to protect users privacy and utilization of social network data are challenging. Most of developer developed privacy models such as K-anonymity for protecting node or vertex reidentification in structure information. Users privacy models get forced by other user, if a group of node largely share the same sensitive labels then other users easily find out one’s data ,so that structure anonymization method is not purely protected. There are some previous approaches such as edge editing or node clustering .Here structural information as well as sensitive labels of individuals get considered using K- degree l-deversityanonymity model. The new approach in anonymization methodology is adding noise nodes. By considering the least distortion to graph properties,the development of new algorithm using noise nodes into original graph. Most important it will provide an analysis of no.of noise nodes added and their impact on important graph property. Keywords-Anonymization,Noise node,KDLD __________________________________________________*****_________________________________________________ I. Introduction The use of social network sites goes on increasing such as facebook ,twitter and linkedin .By using this, users find that they can obtain more and more useful information such as the user performance, private growth, dispersal of disease etc. It is also important that users private data should not get disclose. Thus, how to protect individual privacy and at the same timepreserve the utility of social network data becomes a challenging.Here consider a graph model where each vertex in the graph is associated with a sensitive label. A variety of privacy models as well as anonymization algorithms havebeen developed (e.g.kanonymity,l-diversity,t-closeness). In tabular microdata, some ofthe nonsensitiveattributes, called quasi identifers, can be used to reidentifyusers data andtheir sensitive attributes or information. When circulating social network data, graph structures are alsoissued with corresponding socialrelationships. A structure attack is an attack that uses the structure information or data, that is the degree and the subgraph of a node, to recognize the node. To prevent structure attacks,a published graph should fulfill k-anonymity. The aim is to publish a social graph, which always has minimum k candidates in different attack scenarios in order to protect privacy. A k-degree anonymity model is used to prevent degree attacks. A graph is k-degree anonymous if and only if for any node inthis graph, there exist at least k -1 other nodes with thesame degree. If an opponent knows that one person has three friends in the graph, he can directly know that node 2 is that person and the related attributes of node 2 are discovered. k-degree anonymity can be used to inhibit such structure attacks. Though, in many applications, a social network where each node has sensitive attributes should be circulated. For example, a graph may contain the user salaries which are sensitive label. In this case, only k-degree is not sufficient to prevent the inference of sensitive attributes of individuals. The l-diversity should be adopted for graphs. In this work, selecting the degree-attack, one of the famous attacks methods to show how to design mechanisms of protecting both identities and sensitive labels. Current approaches for protecting graph privacy can be classified into two categories: clustering[7] and edge editing. The method clustering is to merge a subgraph to form one super node, which is inappropriate for sensitive labeled graphs after theyget merged into one super node, the node-label relations have been vanished. Edge editing methods keep the nodes as it is and only add/delete/swap edges. However, edge editing may largely destroy the characteristics of the graph. The distance characteristics get changed substantially by connecting two faraway nodes or deleting the bridge link between two communities in the edge editing method. Miningover these data might get the wrong conclusion about how the salaries are distributed in the the world. Therefore, solely relying on edge editing may not be a good solution to preserve data utility[1]. While considering the above problem, in this work the basic idea is to maintain important graph properties, like distances between nodes by adding certain “noise” nodes into a graph. According to noise adding concept will concern the following observation.Small degree vertices in the graph are used to hide added noise nodes from being reidentified for that purpose widely used Power Law distribution to satisfy social