Further exploring the tools that my Digital Public Humanities (DPH) course has to offer, I took a dive into the world of geographical information systems (GIS) to better my understanding of how datasets can be analyzed and reconstructed into visualizations that facilitate research through computational means. In this case, I used the kepler.gl geospatial analysis tool to plot a set of data pertaining to interviews conducted with former slaves in the United States, similar documents as to what was used in my last post.
Through this tool, I was able to define this dataset visually by inserting it onto a map and using the different features to identify details of the information that allowed for varying interpretations of what resulted from the manipulation. In this post, I will provide a brief guide on how to begin using this program and then my conclusions from the experiments I ran with the dataset and what I was able to discover.
This tool that I used for my assignment (a demo version) allows users to upload files containing organized data that can then be displayed across a world map. The main interface is the map itself with a sidebar of options on the left. The top right corner contains buttons to open a legend (that changes depending on the settings applied to the map), a 3D inversion of the map, and a split screen function to compare maps with different layered settings.
Under the name of the tool on the sidebar are four options. The first is called “Layers.” This section is where data is added and then insert onto the map. So this is where I included the dataset of the ex-slave interviews. Once data has been inserted, the user can add layers to the map. The layers are displays of the data in the form of markers, lines, colors, and shape (as it pertains to the size of the markers). This is the function that plots data. When conducting my assignment, I used different markers to reveal different information. The “point” marker pinpoints the exact location of a line from the dataset (an interview, in my case) according to the coordinates provided in the set. Other display markers are good to describe proximity of plotted points (such as the heat and cluster markers) and others are good for describing relationships between points (such as the arc and line markers). Under this function, multiple layers can be created that plot different sets of data, which is done through the “Add Data” and “Add Layers” buttons. I used this function to add a second dataset of interviews. Doing so allowed me to distinguish between the sets by using different colors for my layers and this presented a different perspective of the data than the previous functions, a perspective that focused more on relationships between the data.
The second option is the “Filter” selection. This allows for further distinguishing of individual lines of data within the set. In my case, I used it to categorize the “type” of slave the interviewees were, either a “house slave” or “field slave.” The filter thus helps to refine the data. Once a set has been entered into the layers section, categories and features become available under “Filter” that corresponds to the metadata in the set. This filter can also add further analysis tools to display certain information. A “time” tool could be utilized because the dataset contained dates and times of when interviews were conducted. Adding this filter allowed me to use a scale to select spans of time that plotted when interviews occurred on the map, which would then fade away as the time span carried on down the timeline.
The third option is called “Interactions.” I didn’t necessarily use this function for my project, but this section allows the user to define the interactive nature of the plotted points of data. For example, the user can edit the information that appears in the data tip information box that appears when a plot is selected.
The last option in this toolbar is called “Base map.” This is where the user can define the backdrop of the data–the map itself. The map can be changed to two variations of either dark or light, which can make the data more readable, and features of the landscape in the map can be altered, such as removing streets and waterways. The user may even add different map styles that can better suit the intended dataset. I used this setting to make my plotted data more clear against the map and remove items that might clutter the information, such as the streets.
There are also three options next to the name of the program in the left sidebar. Two of these are for reporting bugs and a user guide to the program. The third option is an export button and this is how the user may capture the information presented by the map. The user may export an image of the map to show plotted data, export the data itself, export a file version of the map to embed onto other sites, or save the map that is being worked on.
Various other interactions between features I used were the aforementioned top right corner options. When plotting my data through the Layers option, for example, using the 3D model or split screen function was an option to enhance the plotted information. I was also able to layer my…layers…to compare how they rendered, for example, the same information through different markers. For the record, I preferred the heat marker over the cluster marker. Here’s why.
What I Learned
Though the datasets I was working with were basic and minimal compared to the projects we’ve been exploring in this course, they provided enough structure for this program to really experience the functions and demonstrate how geospatial information can yield very nuanced interpretations when the data is manipulated to present novel perspectives.
For example, as mentioned above, I preferred the heat marker plot points over the cluster marker. This is because while they were describing the same information in similar manners, the very appearance of the markers influenced how the information was interpreted. The cluster marker grouped nearby points together and displayed one large circle that mapped the proximity of these points in a region of the geographical area. While this grouping effect is the same parameter for the heat marker, the cluster marker is a blanket rigid shape that doesn’t accurately depict the density of information that might be nearby each other. It simply displays that they are in the same general locality. Conversely, the heat marker is still grouping nearby data together to depict the density in regional localities, but it assumes an amorphous state that contorts according to the proximity of plotted points and relies on a brighter coloring to depict the density, or amount of, interviews in the locality. Thus, it provides a much more accurate picture for the geospatial data as it pertains to specific locations while still providing the desired information of grouped points described by density.
Another interesting point I learned from the operating of this data is how displaying the relationships between points can suggest different interpretations and spur on more research questions. When layering the plotted data to connect the interview locations to the places where the interviewees were slaves, it showed a direct line from where they were, geographically speaking, to where they are now (at the time of the interview). This suggests to me that this tool can be used to track where certain interviewees originated from and where they ended up, perhaps indicating a correlation between the points. Did people who were slaves in one region migrate to another intentionally and for what reason(s)? Were they more likely to pick one place over others?
And a final example was how the time filter altered how I perceived this interviews happening. When all plotted onto the map without the time filter, the data is presented in a very static way. The information is there and one could easily forget that these interviews were conducted over the better part of a year, during different times of this year, and in different frequencies. The time filter addresses all these circumstances to show that these interviews did not all happen at once, that certain periods saw more interviews occurring, and different locations had higher frequencies of interviews compared to others.
This assignment and tool reinforced the notion that was echoed throughout our provided readings: these tools, whether it be the mapping tools or otherwise, are not meant to be the “end” of the data gathered. These tools are meant to conduct research and assist in creating narratives that explain what is being mapped by making it possible to reimagine how this data can be presented through computational manipulation. Though the analog text we’re all accustomed to is often the starting point for these projects and the data these tools utilizes, their 2D presentation has its limits and that much more can be discovered when brought under a different light. Experiences are connected to time and space. Locations are created by individuals. And the geography of those areas tells a story.
When creating my mapping model, I went through several variations to fulfill the requirements of my assignment. Here is one of the variations I created that shows arc lines connecting the locations of the interviews that were conducted in Alabama to the locations where the interviewees were formerly enslaved. The red dots with the pink lines are where the interviews occurred and the black dots with the gray lines are where the interviewees were formerly enslaved.