Papri Mani Int. Journal of Engineering Research and Applications www.ijera.com ISSN : 2248-9622, Vol. 5, Issue 6, ( Part -4) June 2015, pp.38-42 www.ijera.com 38 | Page A Brief Survey on Vertex and Label Anonymization Techniques of Online Social Network Data Papri Mani*, Munmun Bhattacharya** *(Department of Information Technology, Jadavpur University, Kolkata-98) ** (Department of Information Technology, Jadavpur University, Kolkata-98) ABSTRACT With more and more people joining different online social networking (OSN) services every day, the archives of the OSN service providers are increasing drastically. This great amount of personal information is then shared by the service providers with different third parties, which raises a serious concern in preserving privacy of the individuals. For the last few years many work have been done to innovate new techniques, called anonyization techniques, to protect privacy in social network data publishing. In this paper we briefly discuss and categorize vertex and label anonymization techniques which prevent disclosure of individual identities and sensitive information about those identities. We also categorize attributes, attacks and privacy breaches in online social networks. Keywords – automorphism, equivalence class, isomorphism, k-anonymization, social network graph I. INTRODUCTION Online Social Networks are an inseparable part of modern life. They satiate the need and desire of an individual to connect to the rest of the world. A Social Network is represented as a graph where the vertices or nodes represent different real world entities (such as people, organizations or groups) [1] and the edges represent relationships (such as friend, family or colleague) among those entities. People use these Social Networking platforms to connect to their family, friends and colleagues and share their personal views and information with them. So Social Network service providers collect a great amount of private information in their databases. They often share these data with third parties like advertising partners (to get targeted advertisements), application developers and academic researchers. But privacy is a major concern while publishing these data for analysis as an adversary can re-identify a vertex (i.e. an individual), an edge or labels (or attributes) of a vertex using those published data and some background knowledge. In order to stop these privacy breaches many anonymization techniques are adapted while publishing the Social Network data. In this paper we classify the vertex and label anonymization techniques and analyze which kind of privacy attack they prevent. The rest of the paper is organized as follows: in section II we discuss definitions and notations of a few important terms. In section III and IV we discuss classifications of attributes and privacy breaches in social network data respectively. Section V contains the types of privacy attacks on published social network data. Section VI comprises of categorization of vertex and label anonymization techniques. In section VII we discuss related works. And finally we conclude in section VIII. II. DEFINITION AND NOTATION In this section we discuss the definitions and notations of a few terms which are frequently used in rest of the paper. Definition 1. Social Network Graph: a social network graph can be defined as G (V, E, σ, λ), where V is the set of vertices, and each vertex represents an individual in the social network. E ⊆ VⅹV is the set of edges (relationships) between vertices, σ is a set of labels that vertices have. λ: V  σ maps vertices to their labels [9]. We use vertex and node interchangeably throughout the paper. Definition 2. Equivalence Class: equivalence class of an anonymized table data is a set of records that have the same values for the quasi-identifiers [6]. But an equivalence class in a social network graph can be defined in terms of quasi-identifiers values or vertex degree or neighbourhood knowledge or a combination of them. Definition 3. k-Anonymous Graph: A graph G (V, E) is said to be k-anonymous if for each vertex v ∊ V there exist at least other k-1 vertices which have either same quasi-identifiers or degree or neighbourhood knowledge as that of v’s (i.e. the graph can be divided into a number of equivalence classes having at least k number of vertices each). The process of making a graph k-anonymous is known as k-anonymization. Definition 4. Graph Isomorphism: Let G = (V, E) and G = (V, E) be two graphs where |V| = |V|. G RESEARCH ARTICLE OPEN ACCESS