twitter-sentiment-analysis-tutorial-201107

twitter-sentiment-analysis-tutorial-201107 copied to clipboard

INTRODUCTION

This project contains all the code necessary to reproduce the analysis
presented in my R tutorial on mining Twitter for airline sentiment:

	http://jeffreybreen.wordpress.com/2011/07/04/twitter-text-mining-r-slides/

CONTENTS

data/
	opinion-lexicon-English/	- Hu & Liu's opinion lexicon
		negative-words.txt
		NLP-handbook-sentiment-analysis.pdf
		positive-words.txt
		source.txt
		source.webloc

	acsi.df.RData				- airline customer satisfaction scores
	acsi.raw.df.RData			- scraped from theacsi.org web site
	
output/							- output files, mainly graphics
	qplot_delta_hist.pdf
	twitter_acsi_comparison_with_fit.pdf
	twitter_acsi_comparison.pdf
	twitter_score_histograms.pdf

R/								- R source code
	0_start.R
	1_load.R
	2_run.R
	scrape.R
	sentiment.R
	
LICENSE							- Copyright per Apache 2.0 license

README 							- this file

INSTRUCTIONS

In order to run the analysis, start R from this project's root directory or it with setwd()

1. Load the prerequisite packages, our score.sentiment() function, and some environment variables with the "0_start.R" script:

    > source("R/0_start.R")
   

2. This distribution does not ship with any data from Twitter, so you will need to collect your own the first time you attempt to run this package.

   To collect data from Twitter, simply execute the "scrape.R" script:

    > source("R/scrape.R")
   

   This script caches your collected tweets to the data/ directory, so you only need to run this step once.

3. Load the Twitter data, opinion lexicon, and ACSI results from disk:

   > source("R/1_load.R")
   

   If this is your first time running this code, and you have not followed Step 2 to collect your own Twitter data, you will see this error message:

   Error: Tweets not found on disk -- source('R/scrape.R') to scrape Twitter first

4. Run the analysis:

   > source("R/2_run.R")
   

   Progress messages will be displayed on the console and all generated graphics will be displayed and saved as PDFs in the output/ directory.

Jeffrey Oliver Breen [email protected] July 2011