Project files:
DA\results:
- Batch_3727145_batch_results.csv – original Mturk output csv.
- filtered_results_with_zscores.csv – the filtered results with z-scores.
- controls_df.csv – control sentences data only.
- sentences_mistakes_scores.csv – sentences as a vector of NUCLE mistakes with z-scores.
- sentences_mistakes_scores_errant.csv – sentences as a vector of ERRANT mistakes with z-scores.
- mistakes_weights.csv – more statistic information about NUCLE weights.
- mistakes_weights_errant.csv - more statistic information about ERRANT weights.
- bootstrap.csv – 10,000 iterations bootstrap results
- bootstrap_errant.csv - 10,000 iterations bootstrap results on ERRANT
- ranks.csv - 10,000 iterations bootstrap mistakes ranking
- ranks_errant.csv - 10,000 iterations bootstrap ERRANT mistakes ranking
- graphs – graphs folder.
NUCLE\my_NUCLE_parser:
-
my_parser.py – this file parse NUCLE corpus into several databases (regular, perfect and control sentences), according to different filters that serves to create the MTurk csv file.
-
batchCreator.py – python script that write hard-coded JS script for MTurk
-
results_processing.py – main results processing file, including data filtering and re-formatting.
-
results_analysis.py – main results analysis file, create different data sets, and plot the results (imported to results_processing.py and being used by it).
NUCLE\to_Mturk:
-
c_sentences.csv, c_sentences.txt – project control sentences.
-
m_sentences.csv, m_sentences.txt – project mistake sentences – sentence that has been evaluated by one worker only.
-
p_sentences.csv, p_sentences.txt - project perfect sentences – sentences without mistakes.
-
mTurk_csv.csv – final csv to be uploaded to MTurk.