Odysseus

Online social networking websites such as Facebook, Bebo and myriad others are increasingly used purely for recreation by users idly visiting and re-visiting their friend’s pages. We see an opportunity here for a new type of presentation which enables users to spend this time more productively by giving them the requisite tools and information necessary to forge new connections and actively maintain existing ones. We envision a new paradigm of presenting social connectedness, to transform a social network website – which is primarily geared towards entertainment – into a social network application, which aids users in recognising structure in their own network, identifying problems, strengthening latent ties and improving the quality of their social connections.

The goal of the project is to develop a web-based application that does more than present a flat list of pro- file pages. If we abandon the traditional presentation of isolated profile pages which are loaded individu- ally, and instead present a “social space” with a representation of the user’s profile in the centre of the screen and their contacts arrayed visually around it, we can expose the richness of further graph-theoretic measures to the user in an intuitive way. Zooming and panning controls let users explore the relationships between themselves and their friends, but also their friends with each other.

For example, many sites now present a panel enumerating the friends you have in common with the person whose profile you’re currently viewing. Though this gives an indication of your level of interconnectedness, it is limited in describing the more complex concepts of who knows who in your network. By assigning a simple geometry to the network the extent of these connections can be better conveyed. The two- dimensional exploration of this space should be optimised to work on devices like the iPhone. Secondly, in traditional social networking websites, all edges in the network are uniformly weighted – their value is either zero or one. In this new system, each edge has an associated weight, to represent the dy- namic aspect of social interaction. As time goes by, each friend in your network will drift away from your central profile by a certain distance. This can be represented visually in the application, which encourages users to contact their friends regularly to stay in good standing. It also allows them to gain an overview of the general health of their network of friends, classify contacts, and identify problems as they happen (and thus avoid the feeling of “I haven’t talked to them in ages”). As this distance can be seen on both sides of the relationship, both parties can work to fix the disconnect. A beneficial side-effect of this “proximity” met- ric is that it could be enforced on the server so that different privacy settings are applied to contacts which are a different distance out from the user, so that only literally “close” friends would be able to view a user’s pictures, for example. This provides a novel fine-grained approach to privacy.

As the extent of a relationship between two friends is poorly represented by the frequency of their interac- tions solely through any one social networking website, our application will also be able to parse and as- similate information from other traces of interaction. A background client process on the user’s machine can monitor email and instant message communication (and SMS and phone call logs) between users of the network and feed the relevant information into the system. It would even be possible to collect co-location information from a positioning system like Ubisense, so that the relationships between members of a re- search group could be plotted in the system (which we see as an ideal case study).

The application must strike a balance between the user-friendliness of entertainment-focused websites with the theoretical underpinnings and utility of a research project.

Supervisors and Mentors:
Aaron Quigley, Ross Shannon.