train.res file

[anirarrow]

Hi,
I downloaded the train/dev/test.res files of AG News corpus as indicated in your repo. However, I cannot understand the structure of the file. It looks like a simple parse tree, but does not resemble the standard parse trees that we can obtain from the parsers available. The AG News' main website also does not contain these .res files.

Could you please tell me how the preprocessing was done, i.e., how to convert the text into this format of parse trees? Any help would be appreciated.

Regards