Jennifer Chang - Topic Modeling

I had some problems downloading the Topic Modeling Tool at first, but I finally got it to work. I chose Leo Tolstoy's War and Peace because it's notoriously long, and I wanted to see what topics the tool would pick up out of a novel that's supposedly one of the most important works of literature ever written (I have never read it, but I was hoping this would serve as inspiration/motivation to). I admit I still didn't have a great idea of what the book was about after I finished running it through the program:



The results weren't very conclusive, but all of the words were one I'd expect from a Russian novel about war. I went into "Advanced" settings and changed the number of iterations to 100, but kept the number of topics at 10.



This was an interesting change, as it picked up on words that were most frequently used, but not necessarily important (i.e., "called"). I noticed that it read more like a story when scanning the list of revised topics.



Overall, I'm not entirely sure what to gain from topic modeling, but it was still interesting to see how the computer chose to quantify what is a relatively abstract set of information.