How it Works

Our service for Social Network Analysis runs over the web, making it easy for people to view and build their personal networks, and making it possible to use it as a collaborate vehicle.

Building the network

Information on connections can be obtained from a variety of sources, depending on the objectives and what is available:

Billing records: for a law firm, we performed an automated analysis of project billing records, to determine who worked in common sectors
E-mail traffic: e-mail and messaging traffic can be automatically searched, creating connections based on frequency of communication over time
Collaboration: for a leading management journal, we analysed co-authorship of articles to identify groups of experts in key subject areas, and mapped these to find out which authors tended to collaborate with whom
Group membership: membership of groups or committees within or outside of an organization can be used to identify and map connections between people
Interviews: discussions and questionnaires can be used to identify who works with whom, who seeks advice from whom, and other types of connections
Combinations: networks can be built by combining the above sources, for a richer picture

Making it visual

Our tool keeps people and their relationships in a database, from which the site generates diagrams showing all or a subset of people or groups, with lines representing their connections.

These sites are highly visual. Icons, which represent people or groups, can be chosen to represent any number of attributes, such as location, organization, role, expertise, etc., and you can switch between different icon views by selecting from a drop-down list. Different kinds of connections can also be represented (e.g., friendship, reporting lines, project collaboration, co-authorship, geographic proximity, etc.), and these can be represented on diagrams by colour-coded lines.

These diagrams are also highly interactive: you can actually move the icons around by dragging them with the mouse, or hide people, or add connections by clicking and dragging to create lines directly on the diagram. The system then recalculates key social networking metrics to reflect the impact of the change on the strength of the network or on potential collaboration.

In the example illustrated here (click on small image to see larger view), Armstrong has created a diagram with himself at the center, and expanded each of his contacts to see one level removed.

Users can generate and save any number of diagrams under descriptive names, marking them as public if they wish to share them with others (otherwise, they remain private).

Building and exploring your own network

Unlike typical Social Network Analysis, which perform analysis on a pre-collected data set, our approach is user centric: it emphasises the ability for each individual to view, analyse, and monitor their network on a continuous basis. It does this by using the power for the web to give each person their own virtual web site, showing diagrams of their network, and allowing them to add or change people and connections.

While data can be loaded in one time at the beginning of a project (from corporate databases, for example), the database then becomes "live", allowing you to monitor the performance and characteristics of the network as it evolves.

As part of the user-centric bias, we are developing a dashboard that ever user will see upon login (preliminary example at the right, please click on it to see larger version). This dashboard screen will contain summary information about that user's own network, including the people they are connected to, recent activity, and so on. The underlying data is of course part of an aggregate database, from which aggregate views of network are constructed.

Analysing the network

Social Network Analysis dates back to the 1930s, and increasingly sophisticate analyses have been developed to quickly get at the essence of how the network is shaped: is there any clustering, are certain people crucial nodes through which others connect, how connected is everybody to everybody else, and so on. These questions are not academic--they are key issues that need to be understood and addressed for your organization to do its best. Our work supports these analyses in two ways, visual and analytical.

Visual analysis lets you generate, view, and manipulate diagrams of connections, such as that shown at the right (click for larger view). In this example, an academic community has been separated into three clusters, representing three campuses. Diagrams are very useful, and make it possible to see patterns that analytic measures often miss. These diagrams enhance this ability, by allowing you to move shapes around and choose icons to represent different characteristics of the nodes (campus in this example).

Analytic tools use mathematical algorithms from the fields of graph theory and matrix algebra to automatically come up with summary measures that describe an individual, a whole network, or just part of it. They are a useful complement to visual inspection of diagrams.