A network can be defined as a collection of nodes which can be things like products, people and even texts connected by relationships or associations of some kind which are known as edges. A node is an individual point or item in our network and edge is the connection between the nodes. A difference between edges can be seen through thickness, the thicker the edge the more closely associated the two nodes are. Each node can have multiple edges, thickness helps distinguish between the relationships of each node. They can be directed or undirected. An example of directed in real life can be displayed by friendship, you claim a person as a friend, but the person pretends like they don’t know you. Like in instagram if you follow someone they don’t have to follow you back. Then undirected, which is our focus in class, shows no value between the relationship of two nodes. And an example of undirected in real life can be displayed by the use of facebook, if you follow someone, they follow you back but if you unfollow them, they’ll unfollow you as well. A network analysis is similar to that of a cluster analysis because they both use the data of MFW. A network analysis just emphasizes on the CSV file and is different in how the visualization is being presented to us.
I did network analysis on my corpus using Palladio Stanford link. Palladio helps us visualize complex data, with Palladio I was able to copy and paste my CSV file. This CSV file I obtained the same methods as if we were doing a cluster analysis. And in a folder a csv. file was created. I opened it in excel and all it had were sources, targets, weight and type. The sources and targets were nodes and the weight was the edges which contained numbers that told us the higher the number the more closely associated the two texts were which would be displayed by its thickness in edges on a two-dimensional graph. The type in our csv. file is undirected because that’s what we will currently be working with. There were a few hiccups working with this file because commas were very important, because of their placement within the file and how they translate when copy and pasted into Palladio. Some of my texts had commas within the text name and I could have used automator to fix it but I fixed it manually because it was only two texts that was shifting everything and splitting my text names into different columns. An easier way that to make this network analysis that was taught to us at end of use struggling to fix our texts was to us Rstudio. Similar to downloading the stylo package and choosing the stylo option, there was another option for the network analysis called stylo.network which immediately gave us results.
This image was achieved with copying and pasting the csv. file in Stanfords Palladio. And the following image was achieved simply by using R studio.
I noticed that the graph showed a connection between two authors and left an author out on the side alone. To me this didn’t shock me, I expected it because I’ve seen similar results in oppose and zeta where Mark Twain and Charles Dickens were closely associated with one another most likely because of the difference in writing based on gender. The other author that I used was a females author named Jane Austen, again gender, according to the graph and previous analysis, is a big influence in how my texts are being classified. I kind of expected for there to be a connection with Jane Austen and Charles Dickens only because I believed that regional/ cultural writing differences would be influential as they’re both English writers in contrast to Mark Twain who is an American writer. I thought the graph was really good at showing a visualization of how closely associated our texts were. Especially the thickness in lines in which you can see between the two authors Mark Twain and Charles Dickens, the edges are there but they aren’t as thick as the other edges that are between texts of made by the same author. I also kind of hoped that some texts from one author connected with a text from another author. Learning network analysis and placing data accordingly wasn’t as difficult as previous tasks. I felt like it was pretty simple to comprehend especially with the comparisons of undirected and directed to the real world.