search

goodmami / penman

PENMAN notation (e.g. AMR) in Python
https://penman.readthedocs.io/
MIT License
139 stars 27 forks source link

Stored reification table outdated from AMR guidelines #122

Closed flipz357 closed 10 months ago

flipz357 commented 11 months ago

There has been one or two corrections in the reification table of the guidelines, by Jonas Heinecke and me (e.g., see :poss relation where arg0 and arg1 are switched). Maybe it would be good to write a function that crawls always the latest amr guidelines and tries extract reification rule table, and only then if it doesn't work uses a stored table (and if it does work, potentially issue a warning if the crawled version and the version in store are different)?

I'm kinda having this problem also in Smatchpp (for now I took to manually update the table). I can make a pull request in case I find a neat solution, or maybe you have a better suggestion.

goodmami commented 11 months ago

There has been one or two corrections in the reification table of the guidelines, by Jonas Heinecke and me (e.g., see :poss relation where arg0 and arg1 are switched).

Thanks for the heads up, @flipz357. I think this is the relevant context: https://github.com/amrisi/amr-guidelines/issues/261

Maybe it would be good to write a function that crawls always the latest amr guidelines and tries extract reification rule table, and only then if it doesn't work uses a stored table (and if it does work, potentially issue a warning if the crawled version and the version in store are different)?

No, at least not in the library itself. It might be useful to do something like this in a GitHub action on a pull request. But I think this will happen infrequently enough that manual intervention is enough.

flipz357 commented 11 months ago

Yes, you are right, it doesn't happen frequently, so manual should be fine.