Sentiment Analysis: Approach

Approach

I will first reproduce the main Chapter 2 sentiment analysis example from Text Mining with R. The example changes the Jane Austen text into one-word-per-row format, joins the words to the Bing sentiment lexicon, and compares positive and negative word counts across sections of the novels.

I will extend the example by analyzing Moby Dick from Project Gutenberg. I will use the Bing lexicon again so I can compare the new corpus with the original example. I will also add the VADER lexicon and compare its result with Bing. I chose these two lexicons because they score words differently, which gives me a way to check how much the lexicon changes the result. I will present the comparison in tables and color-blind-friendly charts. I will also explain that a word-level method does not fully understand context, negation, or sarcasm.

Sources

• Hutto, C. J., & Gilbert, E. (2014). VADER: A parsimonious rule-based model for sentiment analysis of social media text. Proceedings of the International AAAI Conference on Web and Social Media, 8(1), 216–225. https://ojs.aaai.org/index.php/ICWSM/article/view/14550
• Melville, H. (1851). Moby-Dick; or, The Whale. Project Gutenberg. https://www.gutenberg.org/ebooks/2701
• Silge, J., & Robinson, D. (2017). Text mining with R: A tidy approach. O’Reilly Media. https://www.tidytextmining.com/sentiment