Using Palladio
Palladio is a web-based platform that allows a user to visualize the connections within a data set. With this network analysis tool, a user can upload a dataset which has been formatted according to Palladio’s specifications and then can very easily choose which attributes of the data to highlight and map. The more attributes you have for the nodes in your dataset, the more different visualizations you can create.
The WPA slave narratives, for example, are a wonderful dataset to explore using Palladio. In this case, each interview is a “node,” the kind or type of object that makes up your dataset. But each interview was many different attributes: who was being interviewed, who the interviewer was, the location of the interview, the location where the interviewee had been enslaved, gender of the interviewee and interviewer, age of the interviewee, topics covered in the interview. Of course, this information has to be culled from a dataset and formatted in a way that Palladio can read in order to be mapped. But once that groundwork has been done, the tool allows the data to be explored in many different ways.
To use Palladio, first upload your data and then click “Graph” from the top menu. In the settings box, you will need to both a “Target” and a “Source” to map. These can be any of the attributes that you want to explore in relation to each other. For the WPA narratives, for example, you could choose as the target “Type of Slave” (house or field), and as the source “Location where enslaved” if you wanted to see whether there certain kind of work was done by people enslaved in different locations. The visualization immediately creates a graph that shows the relationships between your chosen categories, with each attribute represented as a small dot with lines connecting those that have a relationship. In the case of the visualization described above, this portrays the categories of “field,” “house,” and “unknown,” each as a circle with lines leading to other circles (the locations where the interviewee was enslaved). This visualization makes it easy to see that the house and field slaves interviewed were largely enslaved in different locations. There are only a few locations where both house and field slaves who had been enslaved there were interviewed. Knowing this might help a researcher who wanted to compare the experience of house and field slaves enslaved in the same place determine which interviews to read more closely.
You have several choices in how the data in your graph is visualized. Clicking “size nodes” makes the size of the circles correspond to the number of data points in that category. You can click or drag any circle to change the shape of your graph. You can also choose whether to highlight the Target or Source information (which presents the circles associated with that information as darker than the others). You can also filter the information that appears on the graph by using the Facet function located at the bottom of the screen, and selecting which dimension of the data you want to filter. The resulting visualizations can easily be downloaded as .svg files by clicking on the download button in the settings box.
What Palladio allows you to discover are the patterns that exist in any given dataset. It highlights which objects or points are most central or important in a particular network and the different links that exist between different objects. What that might reveal specifically depends on the kind of relationships you are mapping in your data. With the slave narrative dataset, Palladio offered a useful way to visualize the locations where interviews were most frequently conducted and where people interviewed at any given location had been enslaved. It offered an easy way to explore the patterns related to specific interviewers—whether they had interviewed more male or female freedpeople, or more house or field slaves, for example. The visualizations are less useful when there are many connections between the different points because the graph becomes so busy it is very hard to read. With the WPA narratives, the visualization of interviews by topics returned a graph where nearly every point was connected to every other. The visualization, in short, demonstrated that for the most part, the different interviews covered most of the same topics.
