“Using Metadata to find Paul Revere” by Kieran Healy uses metadata about a list of the members in each organization to connect the individuals and the organizations.
The data that the author starts from is a simple one by one table with the individual’s name in the rows and the organization’s name in the columns. If the person belongs to that organization, then that block will be filled with 1 and 0 if the individual does not belong to that organization.
With matrix multiplication techniques, they transform this simple table into a matrix of an individual connected to an individual. They make another one that connects two organizations with numbers showing how many people are in those same groups.
Making these two matrices, they made two networks.
These networks show us how the number of connections among the organizations and how the individuals in those organizations are also connected. Without any context, this network analysis gives us a sense of the social structure and possible human interaction.
An interesting (and surprising) outcome of using this relationship between the individual and the groups is that in the individual network, we’re able to single out an individual, in this case, Paul Revere (in the middle of the first network), who seems to be in the center, connected to clusters of other individuals. With this powerful network analysis method, they singled out an individual who seems to be the bridge that ties seemingly separate individuals and organizations.
From this exercise, I realized the power of network analysis and how it can show not only who and what are connected, but also, how everything is connected even if the raw data has no information about that connectivity.
Hi An, I really liked your comment about the power of network analysis! I explored a different project for this exercise, but I also found it amazing how you can end up visualizing a whole social structure, even when you start with raw data on a connection between just two people.
An, I really liked your summary of the method of manipulating matrices that was utilized to create these networks! It’s really neat to think about the different data structures that are at play under the hood to make these network visualizations work in my opinion!