Research Projects on Online Social Networks
Computer Science, Fall 2007
Friday, Dec. 14th, 1pm, BA5256

CSC2231 is a graduate CS course on Online Social Networking Systems. The goal of the course was to familiarize the students with recent research in building large-scale systems that enable mass communication in a social networking manner.

An important component of this course was building and evaluating a system or an application centered around online social networks. The final presentations and demos will be held on Friday December 14th at 1pm in Bahen 5256. Location

Each presentation will last 15 minutes followed by a 5 minute demo and a 10 minute Q&A session. The presentations are open to the public.


1. An Online Social Network-based Recommendation System. Jorge Aranda, Inmar Givoni, Jeremy Handcock, and Danny Tarlow. (ppt, report #1, report #2)
Abstract: We present a social network-based recommendation system that generates recommendations using data from user profiles and social connections between users. Our system uses data from the BoardGameGeek web site, an online social network for board game enthusiasts. Using this system, we demonstrate a prototype web application that allows BoardGameGeek users to obtain board game recommendations based on their social network.

2. Fine Grained Access Control in Online Social Networks. Amin Tootoonchian, Geoff Salmon, and Ahmad Ziad Hatahet. (ppt, report #1, report #2)
Abstract: Online social networks (OSN) have become the primary way many people distribute personal content. However, each OSN may contain only a subset of a user's friends, and the granularity of the available access control may vary. To address these issues we present ImageLock, which lets users control access to their content using social attestations created from their own true social network. ImageLock is intended both to demonstrate to users the advantages of fine-grained access control and to encourage content sharing websites to support a similar scheme.

3. Google Calendar. Maryam Fazel-Zarandi and Lee Chew. (ppt, report #1, report #2)
Abstract: We propose to modify the Google Calendar interface to allow people to publish different views of their calendar to different social networks. For example, a person may want the entries in the calendar between 9-5 on work days to be viewable only by his/her co-workers. Or a person might only want members of the Department of Computer Science to view events pertaining to a lecture series. This interface will leverage an existing social attestation scheme which can establish and verify the relationship between any two people. Our modifications will be integrated with Google Calendar as seamlessly as possible, so as to provide a familiar and unified interface with a very low learning curve.

4. Distributed To-Do Lists Using Online Social Networks. Clifton Forlines, Justin Ho, and Koji Yatani. (ppt, report #1, report #2)
Abstract: We present a prototype that allows friends in an online social network to share "to do" lists to coordinate their shopping efforts. Our system dispatches shopping requests among friends according to their location. We will present a brief demo in which our system requests latte's and mocha's from our friends whenever they visit our local Starbucks coffee shop.

5. Mixxer: Unified Storage and Access Control for Social Networks. Tim Smith and Adam Czajkowski. (ppt, report #1)
Abstract: The emergence of very popular social networking sites, such as MySpace and YouTube, combined with an already existing desire of users to share personal content online, has generated the need for a transparent data management scheme which can not only span all the social networks a person belongs to, but also provide a meaningful way of specifying access control. We present a prototype implementation of such a system, called Mixxer, which allows a user to manage data ranging from generic files, to a Google Calendar schedule, and share it using a social network. In addition, a user can explicitly combine separate identities, from for example Facebook and Gmail, to create a unified repository which is accessible by contacts from both services.

6. Social Networks Swarms in P2P in File Sharing. Wael Aboelsaadat and Sadek Ali. (ppt, report #1)
Abstract: BitTorrent networks are good mediums for sharing large files like popular movies and music. For content that is rare and has a limited audience of interest (i.e., it is unpopular), these networks are inconvenient because the availability of such content is contingent on the provider being connected to the network. In addition, the download of unpopular content does not benefit from swarming because the content is not in demand. An obvious way to improve the availability is to make copies of the content (a.k.a. seeds) in the P2P network. But, without a proportionately large swarm, it may take too long to download the content for seeding. Our proposal is to piggyback the seeding of the unpopular content with the download of popular content in a peer-based exchange of bandwidth for access to desirable content. We present an experimental system that modifies BitTorrent to enable this exchange in a secure and fair way.