technical report

a longitudinal twitter sentiment dataset collected using this tool and its analysis can be found here.

files

methods.py

contains:

• Criterion: a criterion to judge whether a single text sentence falls into any of the possible categories.
  • KeywordCriterion: a type of criterion that uses keywords. keywords can be defined using a dict.
• DistantSupervisor: applies a Criterion to an input file and categorise all tweets in that file, creating a new file for each category. option to also include what evidence based on which it categorised each tweet.

extract_tweets_by_keywords.py

example script for using a KeywordCriterion to categorise tweets into positive and negative based on emoticons and emojis.

analyse_keywords.py

example script for counting how many times each keyword was used as evidence to categorise a tweet (if any) and the how many keywords were found in each categorised tweet.