Within target_extraction.data_types.TargetText.sequence_labels it would be good to have two flags e.g. categories, sentiments where if True then the sequence labels will be like so:
B-POS or
B-LAPTOP#GENERAL
This would roughly follow the same type of structure as the AllenNLP SpanBasedF1Measure
Within target_extraction.data_types.TargetText.sequence_labels it would be good to have two flags e.g. categories, sentiments where if True then the sequence labels will be like so: B-POS or B-LAPTOP#GENERAL
This would roughly follow the same type of structure as the AllenNLP SpanBasedF1Measure