Questions about top-n calculation

[Aminor-z]

In top_scores.py:

def top_scores(probabilities, labels, label_name):
    n_tops = [5, 3, 1]
    if "bugsinpy" in label_name:
        data_split = 15
        window_split = 15
        prob_cutoffs = [0.1, 0.2, 0.3]
    elif "defects4j" in label_name:
        data_split = 15
        window_split = 14
        prob_cutoffs = [0.05, 0.1, 0.2]
    elif "devign" in label_name:
        data_split = 20
        window_split = 50
        prob_cutoffs = [0.35, 0.5, 0.6]

Could you please help explain what the parameters data_split, window_split, and prob_cutoffs represent? I haven't been able to find explanations for them in the relevant papers or training code, so I'm wondering if you might know.