Visualizing 19th Century Social Networks
Those who have followed activities of the Kentucky Historical Society for a while will know about the Civil War Governors of Kentucky (CWGK), a multi-year project to gather documents that are associated with the Kentucky Governor’s Office during the Civil War years and make them available through a searchable database.
So far, CWGK has published more than 10,000 scanned and transcribed documents that tell the stories of thousands of everyday people from across the Commonwealth – many of them stories that would remain unknown today without this project. Bringing them to light is only the beginning. Through a grant from the National Historical Publications & Records Commission, CWGK is working with digital humanities developers at Brumfield Labs in Austin, Texas, to reconstruct a 19th century social network as it appears in these documents.
This starts by identifying all the people, places and organizations in each document – an average of 12 per letter, petition, order or legal document. Then, CWGK researchers write short biographies of each of those individual entities, drawing from all available sources that document every aspect of a person’s family, social, occupational and military activity.
The CWGK network data visualizations, however, will only map the information contained in our documents. We will not network everything that we know. Instead, this is an interesting experiment in mapping the social space described in our texts.
CWGK and Brumfield Labs have entered an exciting stage of work, where the MashBill system that the team has built is now able to generate its first preliminary network graphs. CWGK has only just begun this work, and the data will get more complex and interesting as we link more documents and people together over time.
To establish links, researchers list each relationship they see in a given document, recording which text it appears in and classifying what type of interaction the two people had. (see Relationship Table)
One of the most interesting things about the CWGK visualization system is that it brings together both people and organizations, which shows us not only individuals interacting with one another but also what groups – military units, businesses, churches, civic groups, government agencies – brought them together.
In effect, this is similar to when colleagues ask can if they share any contacts who work with you at your company on LinkedIn or view a list of friends who are in the same sports fan group on Facebook. A less-than-savory example that still illustrates the point is the visualization of the Kentucky State Penitentiary that links the state prison to the inmates who were serving time there and to the guards and prison keepers who appear in CWGK correspondence.
Notice the color differences in the nodes of the visualization. Dark blue dots are people or organizations, and light blue dots are CWGK documents. This is an important and unique feature of CWGK network graphs.
The texts of the documents are more than just the raw material from which we extract information about people’s social lives. The texts help sort and filter information about relationships. Look at this example of a graph of Caroline Dement, an enslaved woman accused of murder in Louisville.
The Penitentiary graph relates individual prisoners to the prison through their documents. Because they were mentioned in a single document each, their connection to the Penitentiary ran through that light blue node. There is a larger set of documents (more than a dozen) about Caroline. People who interacted with Caroline very frequently move inside the ring of light blue document nodes. Willis Levy, the man who enslaved her; his daughter Blanche, whom Caroline was accused of murdering; Gov. James Robinson, who stayed Caroline’s execution; and the Jefferson Circuit Court, where she was tried and sentenced, all move closer to Caroline and have multiple vector lines intersecting a larger number of light colored document nodes.
The graph also does interesting things with the two large petitions signed on Caroline’s behalf. The “dandelion” clusters at 3 and 6 o’clock accurately depict those petitioners’ relationship to Caroline. A number of people signed a document asking for her pardon on grounds of an unfair trial, but their relationship to Caroline was not personal or frequent enough to gravitate any closer or connect to her through more than one text.
These visualizations are just the first draft. CWGK and Brumfield Labs are working on ways to represent weight of relationships (how many times people interact, shown through a darker line connecting them) and a color scheme to show type of relationships (economic, military, familial, etc.).
Notice the colored and weighted vector lines connecting people, documents and organizations in this mock-up of Caroline’s visualization.
These are exciting works-in-progress, and it will take time and effort before the team strikes the perfect balance between aesthetics and interpretation. As the network adds both individuals and connections between them, the system will have to prove itself robust enough to display farmers, seamstresses and governors on an equal footing.
That continuing development work promises to be as rewarding as it will be challenging!