1A_assocations_by_context_term

Can we identify common context tokens (between extracts) that are significantly associated with contentious or non-contentious majority vote sample?

Run:

python3 PIPELINE.py

Requires:

• data.csv from folder

Output:

• majority_vote/majority_vote.csv: a csv of majority vote outcomes by sample;
  • possible options: 1 (contentious), 0 (non-contentious), 0.5 (no overall majority);
• stats/p_response_given_context.csv: a csv of statistics by context word in regards to the strength of its association with contentious or non-contentious majority vote labelled samples;

For Table xx of section 4.3 of the paper: extract a list of those significant context terms with the greatest association with contentious and non-contentious majority vote samples (examining a range of t=min count(unique target terms, context term))

Run:

python3 examine_significant_contentious.py
python3 examine_significant_non_contentious.py

Output:

• contentious_terms_various_t.csv, non_contentious_terms_various_t.csv