Intro

      This graph was created in order to show a relational closeness between every character in the book Les Miserables by Victor Hugo. This book, turned musical, is told from the perspective of characters during the 1832 rebellion in France. Personally, I was drawn to the data set as I once played the character Eponine in a community theater production. Using Gephi, this graph is intended to show how frequently characters came into contact with one another. 

Materials 

      The software used in this lab was Gelphi. The dataset was downloaded from Gelphi’s wiki and important into the software before it was manipulated. 

Process and Discussion 

      Once the data was uploaded to Gelphi, it was simple to use. I was able to run statistics such as Average Degree, Network Diameter, and Graph Density. These statistics would manipulate the visualization in terms of their titles, i.e. Graph Density measured how close the network was to completion. Next I ran several layouts to decide which looked best. This dataset included a lot of information that would need to be labeled, so I landed on using Fruchterman Reingold. It would provide enough space for each character to be labeled, and resembled a dandelion which I liked. After this, I ran the statistic for Modularity to detect an algorithm. With this algorithm I colored the Nodes to indicate clusters of characters who were closer to each other. The more two characters saw each other, the darker blue the node. Finally I changed the degree sizes, labeled each node and changed the labels colors, and exported the visualization.  

       I was largely inspired by others in my Data Viz class who did the same or similar data sets. Nfesta who did the same data set, and Michelle Kung who did a similar one on Harry Potter. It was interesting to see how differently Nfesta’s graph was to mine. They used a lot of color, but had very clear breaking off points of characters.

Reflection       

Gelphi was simple to use once I understood what each term meant for statistics and such. It creates incredibly interesting and beautiful visualizations, but they are sort of a niche visualization. Going forward with the software, it would be interesting to find out if there was a way to filter out characters that have fewer appearances. It would be an interesting project to do a similar visualization but comparing it over episodes of a tv show, and watching how it changes.