Visualizing historical data in the form of networks comes with many problems and benefits. The definition of visualization, as found in Theibault’s article, is “the organization of meaningful information into two or three dimensional spatial form intended to further a systematic inquiry.” Networks are made up of nodes, which are the connection points, and edges, which are the lines between those points. Visualization can be used to identify patterns in data sets and can enhance the presentation of arguments. In this way networks are similar to text mining in that both can be used to identify patterns from a large corpus of information and should be used only as a tool for analyzing sources, not as a methodology in and of itself. In order to be an effective research tool, they need to be information rich, transparent, and accurate. They can be used to show how people move, how information spreads, or where goods travel. Visualizations can serve as either a narrative or an analysis, but the authors of this week’s readings tend to caution against using visualizations as purely narrative structures.
Visualizing networks has many pitfalls, which Theibault, Drucker, and Weingart spend large portions of their articles and blog posts detailing. The network theory of graphs uses mathematical principles that don’t lend themselves to interpreting the experiences of human beings. Drucker maintains that putting statistical information into graphs gives it a semblance of simplicity and legibility, but there is no understanding of or information about the original interpretive framework on which the data of the graph was constructed. Visualizations can be ambiguous and uncertain. Weingart mentions that networks have no memory since the nodes and edges do not contain any information as to how the networks are traversed. Any scholarly work based on quantitative methods need to make clear their research process as well as the process that went into creating the visualization. Since networks are created by algorithms it is important to understand how the algorithms work and to make sure that the mechanics of the visualization are available for anyone to see. By applying your research to a network you are fitting it in to certain standards. How do those standards and the graph affect the integrity of your research? Visualizations has the same problem as all other methods of digital humanities: transparency.
Drucker argues that if the scholarly community wants to use visualizations as a part of humanities scholarship, then all data needs to be reconceived as capta. Data is information that is given, whereas capta is information that is taken actively. According to Drucker, capta is much more representative of how humanities scholars produce knowledge. I would be interested to apply this theory of capta and data to other methodologies utilized in the digital humanities. At first her article was difficult to understand and I had to go back and re-read her main arguments before fully grasping this idea. I had never thought of information input and output in those specific terms, but they do seem to make sense. With the rise of digital humanities, it is important that we learn to differentiate between information that is taken and information that is given.
Friot’s blog post will be helpful for the practicum this week because it was a step-by-step account of instructions on how to use Gephi, a visualization platform. It was also helpful for me to see how I could go about using networks in my own research, starting from the very beginning. Unlike the ngram viewers we used last week, when networking with Gephi you can input your own sources into the program, so I think I will have an easier time understanding how I can use it in my research.
My first thought on examining Mapping the Republic of Letters was that the website should be better organized. For example, when I clicked on Algarotti, I had to look down at the related information to find the visualizations of his travels and network. Shouldn’t that information have been in the body of the site, with the rest of the information pertaining to him? Regardless of this minor quibble, this is an incredibly fascinating site. I particularly liked how the Grand Tour site gave a lot of information as to how they came up with each visualization and the thought process behind them. The page for D’Alembert has the same type of information, and I think any sort of digital humanities should have something in this same vein that explains how the researcher(s) arrived at the visualizations. While they do not explain how the visualizations work, I think it is equally important to detail the beginning stages of such work. Some pages, like Locke’s, haven’t yet been populated with any visualizations, but there is still information as to what the visualizations will be detailing. Mapping the Republic of Letters really helped me to understand how visualizations can be used as a tool to aid historical scholarship. I mean, look at how neat Voltaire’s page is! I especially liked the visualization at the end that networked the correspondents Voltaire and Benjamin Franklin had in common. Mapping the Republic of Letters is a great resource, and provided me with a solid introduction to using visualizations for digital history. I almost think that visualizations are better than text mining, but it does depend on the anticipated end result and what sort of research is being performed. It seems that visualizations provide more information to the viewer, but this could just be a novice assumption.