results.ods: The statistics
forms.csv: A list of the subject forms
easy-forms.txt: A list of the 35 simple forms used in the second experiment, with actions to submit the forms
train_and_test.py: For running the proposed natural-language method
rule_match.py: For running the rule-based method
combine.py: For getting results for the RB+NL-n, RB+NL-m, and RB+NL-b methods
executor.py: A web driver for running the second expeirment
preprocess.py: A library for feature extraction
forms: Offline cache of the subject forms
corpus: Preprocessed corpus from forms
label-all-corpus.json: Labeled topics of all 985 input fields from the 100 subject forms used in the experiments
trial/NUM-training.zip: The first experiment, using NUM% as training data
trial/experiment2.zip: The second experiment, using 7 (20%) of the 35 forms as training data