Summary by Patrick Gossple is an anonymous social network that uses tagging, encryption and onion routing. A user is represented as a Gossple node and each user has a GNet which is a network of semantically close anonymous profiles. Evaluation is done by simulating a network using various datasets from different social networks. The evaluation is based on the quality of the GNets, convergence over time and the bandwidth used by the network. Results appear to be quite interesting, the network uses a low amount of bandwidth. The authors also show an indept example of an application of Gossple as a query expander which uses GNets to improve search results. Summary by Adnan Paper 1: The Gossple Anonymous Social Network. Introduction : Web 2.0 has dramatically changed with the rapid increase of number of people using it. Social media has attracted huge number of people to itself so the numbers are increasing dramatically. To make communication easier tags are used.. This paper proposes improved navigation system between the users and their interests. It purposes associating every user to network of anonymous acquaintances to make the search easier with respect to similar interest. This has been explained using an example of John and alice. Where JohnlivedinUKforalongtimeandnowheislookingforaenglish speaking baby sitter in france. And while doing so he is having so troubles in finding an english speaking baby sitter where as Alice had similar problem once in finding an english speaking teacher assistance. John tried around on facebook among his friends but was not successful. Alice might be an helpful hand in finding the solution for John. But there should be a system to integrate these two people according to their interest or search. Where Alice and Jhon are not friends on social media. Assume that both have tagged their interests in a novel and in a school. First capturing the similar tags with respect to interest is not easy. COntinuing the same example Jhon and Alice has opposite music interest.The fact that Alice and John are identified as acquaintances should not prevent John from benefiting from relevant music information using other acquaintances. Second, discovering social acquaintances might be hampered by the users to publicise their interest information. Not all users would like to share their interests to unknown users or companies. This paper presents Gossple which tried to fix this problem. Basically it uses the concept of proxy and also hidder profile. the Google node or user node generates a request to other user. According to their interests the connection is made without sharing the profile. The profile is encrypted and also to avoid unnecessary bandwidth consumption not ? ?complete profile is shared but only the filtered information is shared.Algorithm GRank is used to rank the web pages according to the described procedure. Evaluation: Gossple is evaluated with a wide?range of Web 2.0 application traces, including Delicious, CiteULike, LastFM and eDonkey, up to 100,000users. By using onion?routing?like techniques Gosspie network is generated with a baseline bandwidth of 15kps. Query expansion application is independently evaluated and shows that the result are better the state of the art search engines.This is achieved in a thrifty manner: 10 acquaintances are enough to achieve effective query expansion in a 50,000?user system, while ex? changing gossip messages and profile digests of approximately 12.9KBytes and 603Bytes respectively. The contributions of this paper are threefold: (i) an anonymous, thrifty gossip protocol ii) a set cosine similarity metric (to compute semantic distances iii) a query expansion application. In the remaining sections of this paper it describes the Gossple Protocol Gossple is a fully decentralized protocol aimed at building and maintaining dynamic communities of anonymous acquaintances. As described earlier overview of challenges is described in this section. On when a user is connected a network is formed. and similar interest are supposed to be found and also to control the traffic generated. So the profiles are linked on the bases of similar interest. to limit the number of matches the number of similar interests is considered. The protocol is divided into two sub protocols (a) Random Peer Sampling protocol (RPS) each node maintains a view of a random subset of network nodes. (b) multi?interest clustering protocol (GNet protocol). ??? Some additional parameters are added here : Time stamp and full profile of nodes that have been chosen as acquaintances. In order to balance the bandwidth usage Bloom filter is used to keep the information bounded according to the profile needs. Gossple Evaluation Evaluation has been done by both simulation (100,000 nodes) and PlanetLab deployment (446 nodes) Gossple at Work: Query Expansion In order to get the accurate result or close to accuracy addition to the query can be done and that is by expanding query by defining more keywords in the query string. Summary by Zahid Social networks is playing a vital role to communicate the people with each others. You can share your ideas and get more knowledge from others but this is main thing that to whom you can share these things. This paper shows a new way to communication in social networks through which you can share your ideas and get communication, called GOOSPLE. Authors used gossip protocol to build a networks of anonymous social acquaintances. In this paper authors give a example of two guys having a problem but opposite direction . But they don¡¯t know each other to share this. Actually one guy name John migrate from USA to Germany and having one kid, he need one baby-sitter who can speak English but he is facing to does not find such kind of baby sitter. In the mean while there is one more guy name Alice who can give services of baby-sitter . Both of them was trying to figure out to find and they used different media near by them. On this way this paper introduce the Goosple that can be use to over come such kind of situation in which anonymous people get together and share their interest in a proper way. Goosple has ability to collect the same data from different user¡¯s profile and provide a way through such people get together having same interest. Goospile get the data from user¡¯s profile and tagged it with those user¡¯s having same interest. Through this way a social networks build and they take benefits even they don¡¯t know each other. In this case of John and Alice Goosple play this role to tagged one¡¯s profile with another and through this way Goosple leverages the very fast that John¡¯s request can benefit from Alice¡¯s tagging profile without decrypt their profiles. Actually Goosple is just like an internet-scale protocol that discovers connections between different kind of users and leverages them to enhance navigation with in the Web 2.0. Goosple has ability to stored the little information and exchanged it with every Goosple user who is associated with a relevant network of anonymous acquaintances. Summary by Ioana The Gossple Anonymous Social Network - Marin Bertier, Davide Frey, Rachid Guerraoui, Anne-Marie Kermarrec and Vincent Leroy (2010) The paper presents a new protocol, Gossple, used to build a decentralized anonymous networks of social aquitances. The goal is to leverage this network within navigation in Web 2.0 applications. Each Gossple node represent a user that has associated a network of aquitances. These can change dynamically, and as a result the profile of a user is automatically updated. Each node retains details about a certain amount of profiles. The selection of these profiles is made using a metric that takes into account the interests of the users, called set cosine similarity. Users send gossips digests of their interest profiles ( the users from their network ) and construct a local view of the network, which is used by each user within Web 2.0 applications. The exchange of the digests is achived with reduced bandwith, due to the fact that users do not echange profiles, but they exchange only Bloom filters of these. The anonimity of the aquitances in the networks is obtain by associating a proxy to each node that gossips for the respective node. The authors use traces from LastFM, Delicious, CiteULike and eDonkey ( ~ 100.000 users ) and PlanetLab deployments ( 446 nodes ) to evaluate Gossple. The traces collected differ in repect with the period considered for collecting the data ( LastFM - Sprint 2008, Delicious - January 2009, CiteULike - October 2008 ), but also in the numbers of users considered. The main findings of the paper in respect to the evaluation process are : - the proposses metric enables the proposed protocol improve the cover of users interests up to 17% (LastFM) and 69%(Delicius) ( in compatison with the state- of-the art ones ) ; - the gossip exchange builds the networks of aquitances in less than 20 cycles; - the bandwidth consumption caused by the exchange is about 15kbps; Questions : 1. Why did the authors choose the four Web 2.0 applications to collect the traces ?