Introduction

Game of Thrones has become one of the most popular television and fantasy book series in pop culture. The expansive universe and elaborate plot lines are a draw for many viewers, but with that comes a huge cast of characters with complicated relationships. It is challenging to keep track of the many characters, their allegiances, and interactions as the story continues. With that in mind I created this visualization to highlight the relationships of the characters based on the third book, ‘A Storm of Swords’.

 

Materials

The data for this project was collected from Network of Thrones, using a CSV dataset containing weighted information regarding the relationships between characters in ‘A Storm of Swords’. The program, Gephi, was utilized to represent the data and create the visualization.

 

Design Choices

In order to create the appropriate visualization for the data I looked at several other visualizations that have attempted to dissect the extended universe of Game of Thrones. I knew that I wanted the visualization to be clean and simple enough to show each character without overwhelming the user. I also wanted to highlight the use of color to separate the major groups of characters.

 

Example 1:

This first example, Network of Thrones, is a visualization utilized by the same dataset as my own, and I based my project off of this source. I liked the general layout and that each node was labelled as the corresponding character with no overlap to ensure that all information was presented clearly. The distinction between the major and minor characters through font and node size was another important component I knew I wanted to include in my own visualization.

 

Example 2:

This example from A Game of Data and GraphQL, takes a deeper look into the universe of Game of Thrones. The visualization depicts several different components of the plot including region, houses, and heirs. The use of color is to signify what information the node depicts (i.e. person, house, region) as opposed to groups of characters as seen in the previous example. This visualization was also featured in the Wall Street Journal as an example of GOT fans becoming obsessed with data.

 

Example 3:

This example from Visualizing Game of Thrones shows connections between characters, their corresponding house, and if/ how they have died. I liked the layout of this visualization with the information forming a circle with the connections confined within the shape. The use of color and scale for the bars also added another layer of information within the visualization. Users are able to clearly identify and understand many different elements of the show within this singular visualization without being overwhelmed.

 

Methods

To create this visualization I imported a CSV file from Network of Thrones. The dataset includes 107 characters from ‘A Storm of Swords’ and are “… joined by 353 integer-weighted edges, in which higher weights correspond to stronger relationships between those characters.” The dataset is comprised of three columns: source, target, and type. Source represents the nodes on the network graph, the target depicts the edges or connections between the nodes, and finally type which denotes the direction of the relationship. The data was already cleaned so it was not necessary to use OpenRefine.

Prior to the lab I watched an introductory video provided by Gephi to become familiar with the program. I also reviewed the class notes and link to the Gephi Quick Start Guide provided on the LMS.

I wanted to create a design that would be clear and simple to understand as a viewer. I wanted to enable users to draw conclusions about character relevance in the overall plot line and understand the strength of relationships between characters within the third book in an intuitive way.

After importing the CSV file into Gephi I browsed the three tabs included in the program: Overview, Data Laboratory, and Preview. I then ran statistics on my data including average degree, network diameter, and modularity. One statistic I found interesting was graph density which was much lower than expected at only 0.062. In class we discussed how many social networks average a graph density of about .20.

For the layout of my visualization I used Forced Atlas 2 and set the nodes by the “Degree Attribute”. To ensure that the nodes would depict an appropriate scale between major and minor characters, I set the minimum to 5 and the maximum to 50. I also prevented overlap to ensure that the nodes would all be visible and clear. I applied color to the nodes to show the different communities included within the visualization. According to Network of Thrones the seven communities are “… the Lannisters and King’s Landing, Robb’s army, Bran and friends, Arya and companions, Jon Snow and the far North, Stannis’s forces, and Daerenys and the exotic people of Essos.” I tried altering the color scheme to be more muted and in line with the mood of the show but found that the palette compromised the clarity of the visualization and ultimately decided to use brighter, more distinct colors.

 

Results

The final result of my project can be found here:

 

I used Forced Atlas 2 for my layout as it depicted the communities and showed a clear hierarchy between major and minor characters. I also increased the edge thickness to 10 to ensure that users would be able to determine the weight of relationships between the nodes.

 

Discussion and Future Directions

Overall I enjoyed working with Gephi, and found the program to effectively create the type of visualization I wanted.

I found the available layouts included in the program to be somewhat limiting, and would like to find an appropriate plugin for additional designs in the future. After exploring some of the plugins available online I was unable to find any that would create a more effective visualization than Forced Atlas 2.

Other options I would be interested in exploring in the future include creating an interactive component to the visualization, adding more datasets to depict other aspects of the universe as shown in Example 3 of the “Design Choices” section, and finally creating an appropriate color-blind friendly color palette.