The script gatherTweets.py reads from Twitter's streaming API, and writes any tweet with an emoticon in it to a file.
The R script scoreEmoticons.R then reads this file, scores each tweet, and searches for the presence of each emoticon (from a predefined list). The score distributions for each emoticon are then plotted.
A more descriptive analysis can be found on my blog Hot Damn, Data! -- The Happiest Emoticons