Today's video class summarizes the approaches that we have taken thus far in the course in order to "read like a computer". It tries to compare them, discuss main differences and suggest some of their limitations. It can be accessed here. The slides are available in our shared drive.
Sentiment Analysis in action (recording)
We will demo some sentiment analysis using the "Detecting Sentiment" notebook in RStudio Cloud.
Sentiment analysis on Twitter Data Regarding 2020 US Elections (Guha)
Assignment 2 allows you to put into action what you have learned about "sentiment analysis".
You will work in pairs (sign up here and list the text you will work with) and choose one text from Project Gutenberg that you are both interested in. Please sign up with someone you have not worked with before. You will work together on both Assignment #2 and the Final Assignment using this same text.
Suggestions:
choose a text that you believe has a characteristic sentiments (horror text - fear; science text - trust).
choose a text that you know already or skim the text to understand what its plot or subject material is.
Run the sentiment analysis notebook with this text and provide an explanation of the visual output, that is, how the computer is reading sentiment. You are, of course, free to "agree" or "disagree" with what the visuals tend to suggest. Use as much context about the book as you have to explain what the computational analysis produces. Remember that there is a big difference between the typical uses of sentiment analysis in business and social media and what you might get for a sustained prose fictional text.
You will co-write this assignment.
Choose your text well, since your final assignment will be to dig deeper into this very same text for the final assignment.
As before, it should be about 1000 words, with appropriate and well described visuals.
For the detailed explanation of the final assignment see this page.