Expected possible value

[q-laurencassidy]

When evaluating the model output using nervaluate.Evaluator, I expected the possible value for the label country to be 1 since there is only one occurrence of country in the gold data. However, the results_per_tag['country']['strict'] output shows possible as 2.

Code:

from nervaluate import Evaluator

labels = ['country', 'postcode', 'city']

true = [[{"label": "postcode", "start": 0, "end": 6, "original_string": "529479"}, {"label": "country", "start": 15, "end": 17, "original_string": "中国"}]]

preds = [[{'label': 'postcode', 'start': 0, 'end': 6, 'original_string': '529479'},
 {'label': 'city', 'start': 9, 'end': 15, 'original_string': '漯河市'},
 {'label': 'country', 'start': 15, 'end': 17, 'original_string': '中国'}]]

evaluator = Evaluator(true, preds, tags=labels)

results, results_per_tag, result_indices, result_indices_by_tag = evaluator.evaluate()

print(results_per_tag['country']['strict'])

Observed output:

{'correct': 1,
 'incorrect': 1,
 'partial': 0,
 'missed': 0,
 'spurious': 0,
 'possible': 2,
 'actual': 2,
 'precision': 0.5,
 'recall': 0.5,
 'f1': 0.5}

Please let me know if there is any misunderstanding in the way possible counts are calculated for each label or if there are other conditions affecting the possible value.