The tag of "AM-ADV" appears to be classified as that of "V".

[smiyawaki0820]

The label of "AM-ADV" (or "C-V") appears to be wrongly classified as the tag of "V" in order to use the condition of "V" in label (judgement for a verb), when I used the script of "CoNLL_to_JSON.py" to create data for the model.

The relevant part of the script is in https://github.com/Heidelberg-NLP/SRL-S2S/blob/master/pre_processing/CoNLL_Annotations.py#L197

[angel-daza]

The cases of AM-ADV and C-V should be caught by one of the first portions of the IF statement [ Namely self.op_lbl_05.match(lbl) ]

[smiyawaki0820]

I am grateful for your answer and as you have indicated such labels seem to be actually caught by the first part of the IF statement.

However, in the case of a label whose span is composed of one word, like (AM-ADV*), would not be caught by the above portions: if self.op_lbl_05.match(lbl) and ")" not in lbl:.

And also, in order to use not search but match function of the methods of regular expressions, this label would not be caught by the second portion capturing a closing label: elif self.cl_lbl_05.match(lbl): (self.cl_lbl_05 is equal to re.compile('\\*\\)').)

For this reason, the label of (AM-ADV*) might be classified as the tag of V.