Introduction to Network Analysis – Gephi

The largest centers of influence are mainly magazines, though there are a few authors in there. Among the top magazines are, of course, Blast, The Little Review, Scribners’, and The Owl. Top authors include Wyndham Lewis and Mercure de France.

 

In terms of least important, it is very difficult to make out specific items due to the density of the graph and sheer amount of items.

Most of this graph was exactly what I would expect, with a few items taking of the majority of connections and a lot of outliers. The graph definitely made this very easy to demonstrate.

Gephi Graphing Lab

Braigen, Reagan, Megan, and Paige

  1.  The Review, whether a topic within a magazine or the magazine itself was a very common link to multiple nodes.  We agreed this, along with the word “the” share the center of the influences in this specific networks.
  2. We came to the conclusion that the magazines seem to be more significant/important within the network than the authors were because in the middle most of the nodes were magazine titles and not authors names.
  3. Some examples of the least connected authors and magazines seem to be Dargan, Tarbell, Elsa, along with The Cosmopolitan, Bangs.  These authors and magazines with minimal connection to the network graph shows us that they do not fares common themes or topics that might be found in other more central works.
  4. Coming in to this lab with never seeing a network like this, there were very little expectations as to how the network would work and what exactly it would do.  The network graph shows value supporting ideas and topic as connected through different magazines along with different authors, however it also brings to light the fact that the author or magazine does not need to be connected to popular topic to exist.

-Word up!

Gephi-lab

  1. The most centered authors seen from the data are Wyndham Lewis, Scribbners, and Charles
  2. Magazines seem to appear more in the center/nodes with larger connections than authors do. Especially when starting at 40 connections and slowly going down, magazines show up more.
  3. When you start looking at nodes with less connections than 10 its start to be too much to evaluate. If you just look at all 1 connections nodes, only 2 of them are connected. Since this seems to just take all words amongst several magazines, A and The dominate whereas Around and Exclusive are outliers.
  4. For only having 2 issues, Blast was one of the top nodes per connections. To no ones surprised ‘The’ and ‘A’ are the top picks when filtering by nodes with the most connections.
  5. Screenshot from 2017-03-22 14-55-28.pngScreenshot from 2017-03-22 14-47-32.png

Gephi Lab

Cox, Flores, Sohl, Zaidi

Gephi is a useful way to visualize networks, but it has some bugs. The questions asked of us for this post are difficult to answer. Gephi breaks up the titles of the journals in the .csv file into individual words – not the whole title. This means that when we select the node “poetry” the nodes it connects us to are “the” and “a.” Ideally, “poetry” would connect to the nodes of all the journals, and then from each journal, it should connect to the various works tagged “poetry.”

gephi

Here is the Yifan Hu layout. Gephi would be a lot more useful if some of its bugs were worked out.

 

Gephi Lab

  1. What authors and magazines are the centers of influence in this network?

    Scribner’s, BLAST, and The Little Review are three magazines play large roles in the graph (because they are the largest nodes). We’ve attached an image of how large the Scribner’s and Little Review magazines are.
    screenshot_144030

  2. Do authors or magazines seem more important here?

    Magazines seem to have the greater importance here, which would make sense considering how many of the different authors and works are associated with them.

  3. In contrast, what magazines or authors are least connected, and what does that tell you about these data?
    Whenever we changed the layout to the Fruchterman Rheingold layout, a number of authors and pieces were pushed to the outside of the graph. Those are the least connected. Here is a visualization:
    screenshot_144446
  4. Does it show you anything unexpected? If so, what is it, and how did the graph help you to notice? If not, what is the value of the network graph as evidence, anyway?

    The most unexpected find was Oreo manufacturers were the least connected in our graph. Additionally, we did not expect the words “a, an, the” to bias the graph as much as the three words did. They were some of the largest players in the graph which caused unnecesary noise in the graph. However, the graph did help us to see what types of pieces are associated with the magazines, such as poetry, essays, and advertisement, etc.

Group: Thomas Littlejohn, Albert Song, Zihao Lu

Literary Maps

“There is a  very simple question about literary maps: what exactly do they do?  What do they do that cannot be done with words, that is; because, if it can be done with words, then maps are superflu­ous.  Take Bakhtin’s essay the chronotope: it is the greatest study ever written on space and narrative and it doesn’t have a  single map.  Carlo Oionisotti’s Geografiae storia della letteratura italiana, the same.  Raymond Williams The Country and the City, the same.  Henri i ESIiaces romanesques du XVllle siecle …  Do maps add anything, to our knowledge of literature?”

I would argue that literary maps are useful. Although they aren’t a crucial necessity, as proven by multiple significant works lacking them, they still provide the reader an extra layer of depth that they may not have received otherwise. Especially for those who choose to study and delve deeper than the average reader, these maps can offer valued insight on topics that could range from other source material as well.

Network Analysis

“The reason weight and direction are particularly important in literary networks is that, whereas the systems studied by network theory have easily thousands or millions of vertices, whose relevance can be directly expressed in the number of connections, plots have usually no more than a few dozens characters; as a consequence, the mere existence of a connection is seldom sufficient to establish a hierarchy, and must be integrated with other measurements.”

There is a related issue in both chronological and spatial analysis of literature that heavily differentiates the two methods. In chronological analysis, when trying to organize text mining and word trends, sometimes these get lost in the multitude of other material in the bodies of work. When there is a large magazine, there may be many other articles or advertisements bogging down overarching technical analysis. There is just a lot of extra,  information. However, in a spatial analysis, it is not hard to find the connections, or “edges” you’ve decided to search for. In fact, there are almost always too many edges apparent in your spatial analysis. If you include all instances of soliloquy, or aside, or dialogue, there will usually be hundreds if not thousands of instances and they are all relevant, albeit to varying degrees. An instance of small talk in passing will rarely compare in weight to extended main character dialogue. The general problem of sifting through massive amounts of info is the same between spatial and chronological analysis, but the infrastructure is the issue.