A. Building Visual Graph of Social Groups Our system produces such a graph that helps users to see each social group and helps the user to make better decisions about his/her application and privacy settings. B. Privacy-Settings To Protect Personal Information Our approach suggests the set of friend lists that users should create, and the friend lists into which they should put each of their current friends based on the identiﬁed social groups. IV. CLUSTERING Methods for clustering have been deeply studied. But our aim is not to study them. Clustering is just one of the steps to achieve our privacy goal. Our main aim is to use the right clustering algorithm for social networks and develop an approach to provide privacy by adapting this clustering algorithm to our application. The clustering in social networks requires grouping users into classes based on their attributes, properties of personal relationship, web page links, spreads of messages and other applications. It is the process of organizing users into groups whose members are similar in some way. Our algorithm is different from other clustering algorithms, and it can dynamically group users in a social network into different classes based on their properties and effectively identify relations among classes. It collects some data which are similar between social network users and are dissimilar to the users belonging to other groups. It creates active cells like network grids and builds 3D visual graph of social groups. The similar structure applied in the algorithm [10, 11] for finding (α, β) clusters has been used in our algorithm. In our application, friends sharing common personal information are the adjacent nodes to α-fraction. On the other hand those friends not sharing common personal information are the adjacent nodes to β-fraction. It is therefore possible to use the social graph of network users as an input to our algorithm. One might ask that what if a friend belongs to more than one group? For example, a user can have a friend from high-school or university that is currently his/her work mate. The overlapping sets or being in more than one group does not cause a problem from the privacy perspective. Our approach groups friends according the common information that a user wants to share with his/her friends. For example, if we just want to share our photos and status with our high-school friends, then we will be showing them a profile where they will only able to see our photos and status. If there are some other friends that we just want to share our photos and status, they will also be in this group. Therefore, it is perfectly normal and possible that a person can be in one or more social groups. The user will show some information in one group and will show different information in another group. In other words, we will limit who sees what. Fig. 1 and 2 show our clustering algorithm and a visual graph of clustering process for the users in a social network, respectively.
...(download the rest of the essay above)