Closed nicholasjhorton closed 7 months ago
I've noticed there are several chapters in Moby Dick which are purely descriptive rather than narrative. My idea is to use some type of topic modeling to cluster sections of text depending on whether they are descriptive, narrative, or some other category I have yet to discover. Sentiment analysis and LDA will probably be useful in doing this.
A larger idea would be to summarize the text based on the apparent themes or character interactions. This could include some measure of character dialogue gathered from interactions between each character. We could use NER to identify people in the text, then measure the proximity between two or more characters based on line number. This method could identify more narrative sections, while the previously-mentioned tools like sentiment analysis and LDA could pick out the themes of descriptive sections of text.
This project would most likely result in a report detailing the exploration of this text with data.
I like this idea: I think that it could complement other literary analyses of Moby Dick.
For next steps can you please open and describe two new issues:
Can you please prep both issues when you have a chance (ideally before class on Tuesday) then close this issue which I've reopened?
The next step for your book package is to propose a substantive analysis that utilizes some technique in text analytics to explore or extract insights.
Your audience can be people new to the book, or people who know it and love it (and can quote from it).
Your deliverable is open-ended: this could be a report, a Shiny app, or something in between.
My hope is that you dream big: we can work together to come up with something that you'll be proud to show off.
Have questions? Have ideas? Please use this issue to share those thoughts.
Finally, please comment to share your proposal (then close the issue to let me know that you are ready for my review).