Summary by Ioana Robust Overlays for Privacy-Preserving Data Dissemination over a Social Graph Abhishek Singh, Guido Urdaneta, Maarten van Steen and Roman Vitenberg The article proposes a solution for building decentralized privacy - preserving communication overlays. One existing solution is using friend-to- friend(F2F) networks that leverage the existence of relationships among users to build communication links. However, this approach does not provide the optimal solution for data dissemination. The paper proposes a solution that provides privacy - preserving data dissemination, by using also trust relationships between users, but improving the robustness of the trust - based overlays by augmenting with additional links. The overlay network satisfies the following: - privacy preserving overlay links; - privacy preserving overlay maintenance; - robustness; The evaluation of the system is conducted using a social graph sample from Facebook and a synthetic churn. The system is analysed in two different scenarios, using a pre-defined set of parameters. The main findings that the systems is able to produce robust overlay and provides fast convergence. Questions : 1. Why is social graph obtained from Facebook data ? 2. Considering the case in which the solution proposed would use trusted graphs that would change over time, how would the system change( complexity, parameters) ? Summary by Zahid Robust overlays for privacy-preserving data disseminationovera social graph Summary : Due to the increase of social networks the user¡¯s data privacy is getting main concern. Each and every data the graph of social network¡¯s users are going higher and higher, like Facebook and Twitter. Facebook receives more than 30 billion shard items every month while Twitter receives more than 55 million tweets each day. Analysis and news shows that free centralized services cannot be trusted to safeguard sensitive online social network data. When one organization is controlling a large amount of data a single sight than the provider the right to reuse user¡¯s data in any way the provider sees fit for different purposes. Freenet pose a strong potential for catering to full fill the decentralized storage and retrieve the data. The main approach in this article is to bootstrap the communication overlay using the social graph. To increase the additional links between pairs of nodes that correspond to the users not related by social ties. These link enhance the privacy preserving routing between pairs of untrusted nodes that can not trust on each other. The main goal of the authors is to produce an overlay protocol that facilitates robust privacy=preserving data dissemination for a group of users that have trust relationships among them. Summary by Sasha Rudan - paper presents an idea of extension of F2F (friend-to-friend) networks in a - wat of introducing more robustness and lower path length - the idea is to use standard trust graph topology as a basic initial - infrastructure of the network, however it is used only to bootstrap - communication overly that is later augmented with additional links - Therefore, communication overlay consists of trusted links and pseudonym - links - trusted links are standard links and pseudonym links are additional links - added on the top of trusted links in a random manner which will increase - randomness of graph, introduce redundancy in paths and shorten path lengths - at the same time authors argues that this will not introduce any security - issues by using non-trusted nodes for communication - solution consists of 3 layers (where top one belongs to application scope) - the bottom one consists of 1) anonymity service: allows nodes to create - privacypreserving links to any destination whose ID is known, 2) pseudonym - service: allows nodes to create pseudonyms and to establish privacy- - preserving links to any destination for which a pseudonym is known. Authors - propose gossiping algorithm for distributing pseudonyms. Pseudonyms have - limited life-time to makes attacks more difficult and support resources - management Summary by Adnan Online Social Networks (OSN) are growing rapidly today and have become the most important aspect in internet community. Social networking websites like Facebook and twitter are growing dramatically. Not only number of people using it but amount the data shared using different social network has increased. This data may be in terms of some pictures or some text information etc. As the number of participants is increasing so is the threat to privacy of the information shared on social media is increasing. In a group of friends a topic is discussed or shared over a social networking website, What is the guarantee that this shared information will not be given to any other person excluding that group members? The debate of centralized and decentralized system continues again and so we get many solutions suggesting decentralized systems. Friend to friend (F2F) is one of such solutions based on decentralized system to work on batter user privacy. It helps in making batter privacy if members in a social webs site like Facebook. For example F2F user¨ªs network cannot find out who else is participating beyond their own cycle of friends. This paper introduces concept of overlays links using the social graph with additional links between pair of nodes that corresponds to users not related ties. Churn is used to make sure that sending data over these links does not violate privacy of the nodes, furthermore mechanism for establishing these links is itself privacy preserving and it does not impose signi?cant overhead, and the resulting overlay becomes robust under churn as well as having a low diameter. It uses pseudonym based routing facility to generate pseudonyms. The trusted overlay links are formed between the trusted nodes and TTL is also considered to make robust overly system. The paper does analyze privacy protection under different scenarios. It also does perform some attacks against the system to check the vulnerability of the systems. Experiments are conduced to check different aspects like the characteristics of the resulting overlay, the overhead of overlay maintenance and the speed of convergence towards a robust overlay. This system produces a service for maintain an overlay for privacy persevering data dissemination for a group of users in a trust based social relationship. Summary by Patrick This paper presents a robust for privacy-preserving data dissemination over a social graph. The idea behind this paper is to create a graph-model where each user is represented by a node in graph, each edge is a connection to another user they trust. The architecture allows for users to connect to users they don't trust through pseudonyms. The pseudonyms can communicate using some external system like email or DHT (Distributed Hash Table). Evaluation is done by simulation and the social graph is constructed by facebook data and by using synthetic churn models. The anonymity and pseudonym services are assumed to exist and that they are ideal and reliable.