Documentation

[cimeister]

It would be great if someone could provide a bit of documentation for this repository! I've started to use it and have had to read the codebase to try to figure out how things work. I would be happy to help in any way I can.

[guyemerson]

Thanks for getting in touch! And sorry we haven't been better about writing documentation. In the short term -- is there anything in particular you're trying to do that I could help with?

If you're happy to help with documentation, that would be fantastic. If you would like to write something, you are welcome to make a pull request.

Have you run into any problems so far?

[cimeister]

Thanks, Guy! So I'm trying to decompose sentences into simpler sentences using DMRS (i.e. The boy named Daniel went to the store -> The boy is named Daniel. The boy went to the store). I was hoping I could use the pydmrs library to make cleaner modifications to DMRS structures, which I could then feed to a generator such as ACE or LKB. Modifying the structure while maintaining a valid form has been the difficult part... Do you have any recommendations within (or outside of) this library?

I can definitely write up some "quick start" notes along with notes about anything else I end up using!

[emm68]

Hi Clara!
The problem with maintaining a valid form is that the valid form depends on the grammar, not the DMRS structure itself. Two rules I can think of are that a) a DMRS has to be fully connected, b) every noun-like thing (with cvarsort x) has to have a quantifier. I can't really tell you more about keeping the modifications clean without knowing more about what kind of operations you're aiming for. That said, I've been messing around with DMRS structures a fair bit for a while now and I'd be happy to share what I've figured out. Perhaps, there is a set of generally useful operations that we could add to pydmrs. I'd be happy to chat about it :) Please feel free to just email me directly.

I'm working on a related problem, but mostly identifying semantically contained fragments, without restoring sentence form (example). Haoyue was an MPhil student who did some work on the restoration of partial DMRS. Section 3 of this paper somewhat describes her approach, but I suspect it doesn't contain enough practical details to be helpful.

[cimeister]

Hi Ewa! I'll email you directly :) I'll keep this issue open as a reminder to myself to work on some documentation...