Proposal Outline

## ################################## ##
## CS 366 - Computational Linguistics ##
## Final Project Proposal             ##
## Jesse Mills & Sam Faber Manning    ##
## YouTube Sentiment Analysis         ##
## ################################## ##

# 1 - Problem and Motivations

Problem -----
- Design an application that scrapes a YouTube video's comments, formats them in a usable way, and then performs sentiment analysis on them, providing the user with various analytics regarding them. 
General Trends ---
- Positivity vs. Negativity
- Polarization (Intensity)
- Sentiment over time
- Sentiment of top-level comments vs. reply comments
- Sentiment of comments re: video vs. comments re: other comments?
- 

Motivation - 

# 2 - Exisiting Work & References
http://home.engineering.iastate.edu/~zambreno/pdf/KriZam13A.pdf
http://www.l3s.de/~siersdorfer/sources/2010/wfp0542-siersdorfer.pdf
https://www.aclweb.org/anthology/P/P14/P14-1118.xhtml

http://ytcomments.klostermann.ca/

The YouTube API has a sentiment analysis demo, but it doesn't seem to utilize an accurate sentiment analysis algorithm.

# 3 - Proposed Solution(s)

# 4 - Work Completed
- Found a scraper
- Wrote a script to POS tag the comment text (only!) from the scraped comments
- Found a tagged corpus of YouTube comments (SenTube)

# 5 - Member Responsibilities

# 6 - Milestones

# a - May 3

# b - May 18 (Minimum)

# c - May 18 (Ideal)


Sentiment Analysis ------
- Negation


Algorithm Steps
1) Scrape Comments -> Store in CSV
2) POS tag the comment field of all the comments
3) Sentiment