utils.py

# Lifted from: https://github.com/shimaokasonse/NFGEC/blob/1171bb4d13269e3614b6216f47cb7090a97e409b/src/evaluate.py

def f1(p,r):
    if r == 0.:
        return 0.
    return 2 * p * r / float( p + r )

def strict(true_and_prediction):
    num_entities = len(true_and_prediction)
    correct_num = 0.
    for true_labels, predicted_labels in true_and_prediction:
        correct_num += set(true_labels) == set(predicted_labels)
    precision = recall = correct_num / num_entities
    return precision, recall, f1( precision, recall)

def loose_macro(true_and_prediction):
    num_entities = len(true_and_prediction)
    p = 0.
    r = 0.
    for true_labels, predicted_labels in true_and_prediction:
        if len(predicted_labels) > 0:
            p += len(set(predicted_labels).intersection(set(true_labels))) / float(len(predicted_labels))
        if len(true_labels):
            r += len(set(predicted_labels).intersection(set(true_labels))) / float(len(true_labels))
    precision = p / num_entities
    recall = r / num_entities
    return precision, recall, f1( precision, recall)

def loose_micro(true_and_prediction):
    num_predicted_labels = 0.
    num_true_labels = 0.
    num_correct_labels = 0.
    for true_labels, predicted_labels in true_and_prediction:
        num_predicted_labels += len(predicted_labels)
        num_true_labels += len(true_labels)
        num_correct_labels += len(set(predicted_labels).intersection(set(true_labels)))
    precision = num_correct_labels / num_predicted_labels
    recall = num_correct_labels / num_true_labels
    return precision, recall, f1( precision, recall)