When parsing sentences containing the token '24/7', the JAMR parser returned AMRs with things like:
:ARG1 24/7
Which results in errors when attempting to further process these AMRs; for example, when running smatch, it returns the error:
Traceback (most recent call last):
File "smatch.py", line 927, in
main(args)
File "smatch.py", line 827, in main
amr1.rename_node(prefix1)
AttributeError: 'NoneType' object has no attribute 'rename_node'
This can be avoided by adding quotation marks in the parser output to treat the problematic token as a string, e.g.
:ARG1 "24/7"
Which appears to be the approach used in gold AMR data.
When parsing sentences containing the token '24/7', the JAMR parser returned AMRs with things like:
:ARG1 24/7
Which results in errors when attempting to further process these AMRs; for example, when running smatch, it returns the error:
Traceback (most recent call last): File "smatch.py", line 927, in
main(args)
File "smatch.py", line 827, in main
amr1.rename_node(prefix1)
AttributeError: 'NoneType' object has no attribute 'rename_node'
This can be avoided by adding quotation marks in the parser output to treat the problematic token as a string, e.g.
:ARG1 "24/7"
Which appears to be the approach used in gold AMR data.