def test_collect_named_entities_same_type_in_sequence():
tags = ['O', 'B-LOC', 'I-LOC', 'B-LOC', 'I-LOC', 'O']
result = collect_named_entities(tags)
expected = [
Entity(e_type='LOC', start_offset=1, end_offset=2),
Entity(e_type='LOC', start_offset=3, end_offset=4)
]
assert result == expected
def test_collect_named_entities_entity_goes_until_last_token():
tags = ['O', 'B-LOC', 'I-LOC', 'B-LOC', 'I-LOC']
result = collect_named_entities(tags)
expected = [
Entity(e_type='LOC', start_offset=1, end_offset=2),
Entity(e_type='LOC', start_offset=3, end_offset=4)
]
assert result == expected
}
# overall results
results = {
'strict': deepcopy(metrics_results),
'ent_type': deepcopy(metrics_results),
'partial': deepcopy(metrics_results),
'exact': deepcopy(metrics_results)
}
# results aggregated by entity type
evaluation_agg_entities_type = {e: deepcopy(results) for e in ['Symptom']}
for true_ents, pred_ents in zip(y_test_, y_pred_):
tmp_results, tmp_agg_results = compute_metrics(
collect_named_entities(true_ents), collect_named_entities(pred_ents),
['Symptom'])
for eval_schema in results.keys():
for metric in metrics_results.keys():
results[eval_schema][metric] += tmp_results[eval_schema][metric]
# Calculate global precision and recall
results = compute_precision_recall_wrapper(results)
# aggregate results by entity type
for e_type in ['Symptom']:
for eval_schema in tmp_agg_results[e_type]:
def test_collect_named_entities_no_entity():
tags = ['O', 'O', 'O', 'O', 'O']
result = collect_named_entities(tags)
expected = []
assert result == expected
def test_collect_named_entities_sequence_has_only_one_entity():
tags = ['B-LOC', 'I-LOC', 'I-LOC']
result = collect_named_entities(tags)
expected = [Entity(e_type='LOC', start_offset=0, end_offset=2)]
assert result == expected