requirements.py

from snorkel.labeling import labeling_function

# Give a meaningful name to each value
FAKE = 1
REAL = 0
ABSTAIN = -1

@labeling_function()
def no_requirements(x: pd.Series):
    return FAKE if x.requirements == "" else ABSTAIN


@labeling_function()
def requirements_less_than_10(x: pd.Series):
    num_words = len(x.requirements.split(" "))
    return FAKE if num_words <= 10 else ABSTAIN


@labeling_function()
def requirements_less_than_20(x: pd.Series):
    num_words = len(x.requirements.split(" "))
    return FAKE if num_words <= 20 else ABSTAIN