Tyler Rinker

The license for the NRC data set in the README asks that the data not be duplicated. https://juliasilge.com/blog/sentiment-lexicons/ Use https://github.com/EmilHvitfeldt/textdata to pull the lexicons and reshape on the fly like...

# Categorized - https://github.com/imsky/wordlists # Bad Words - https://github.com/LDNOOBW/List-of-Dirty-Naughty-Obscene-and-Otherwise-Bad-Words/blob/master/en - https://github.com/reimertz/curse-words/blob/master/index.js - https://github.com/MauriceButler/badwords/blob/master/array.js

Additional lists/dictionaries from discon???

3

https://github.com/trinker/discon

Add a parallel option

5

A parallel option that runs `sentiment` and `sentiment_by` on multiple cores

Include sarcasm in the algorithm

4

- http://www.cs.utah.edu/~huangrh/official-sarcasm-cameraReady-v2.pdf - http://stackoverflow.com/questions/14097388/can-an-algorithm-detect-sarcasm Difficult task for 100% accuracy but there may be key features that are highly correlated with a sarcastic comment that would improve sentiment detection. The idea...

Consider a quotations weight

1

Like questiins...quotes are often not indicative of the speakers sentiment. Consider a way to weight these similar to questions.

https://medium.com/@datancoffee/opinion-analysis-of-text-using-plutchik-5119a80229ea http://examples.yourdictionary.com/examples-of-interjections.html

Possible cleaning function for reimport

5

This would belong in textclean but things that are abbreviated forms like fan vs fanatic: ``` > sentiment(c("He's a nice guy", "can be a jerk. I'm not a fan.")) element_id...

Percents above bars rather than counts

1

Currently counts are returned above the bars. Often percentages are more informative. Could percentages of the total be a feature?

Check package name availability

1

Here's a potential function that checks if a package name is available. It would need to check bioconductor as well but I am uncertain how to do this programatically. ```...