Network graphs are extremely helpful in determining links or connections between objects. In this case, the network graph allows us to visualize which characters from the short story interact with each other. The interactions are shown through edges or the lines connecting the dots or nodes. The font size of the characters’ names as well as the node size indicate who has the most connections, although this can also be determined simply by counting the number of connected edges.
In this graph specifically, we can see that the characters Calvin Stern and Felicia Rhee are isolated and only connected to each other. This is because they are only spoken about in connection to each other in the story, creating a sort of hearsay situation in which there is no direct contact with other characters (or at least from what I could determine in my interpretation of the story).
While the network graph does a good job in showing connections, it comes up short when determining duration or relative importance. For example, Dr. Meyer could have had conversations with a bunch of characters that consisted of “Hi”, “Bye”. While the network graph does show that he is connected to a lot of other characters, it is unable to calculate the relative importance of those connections. Perhaps Calvin Stern and Felicia Rhee are omniscient beings that oversee or run the situation from afar and are only in contact with each other out of necessity. But because they are only talking to each other, the network graph interprets them, and subsequently their relationship, to be of lesser importance.
This shortcoming ultimately comes down to the available kind of data. Since this network graph is based off contact, and contact alone, that is how importance and visual size will be calculated. There are obviously ways to recreate the initial csv to account for these exterior features, but it would require a lot of manual input. For what it is, the network graph can create cheap, simple visuals to highlight basic connectivity concepts. However, if there are more obscure connections and relationships that need to be explored, one must find other ways (whether it be creating multiple graphs or manually adding other features to the backend data) to highlight them.