Deeper nested child annotations about pympi

[dopefishh]

closed by #14

[dopefishh]

Thanks for the feedback. There is no such way I think. However, one could add it.
I haven't been working on this for quite some time so I will have to delve into the matter again and could use all the help to speed things up. I'll try to be back at you asap.

[nikopartanen]

Thanks for reply! Take your time! I usually use R when I work with ELAN files, but if I need to create or populate new tiers I've often relied to pympi anyway. All in all it is very useful package, thanks for great work!

Kind of use cases I would wish to do with pympi are, for example, taking the transcription tier, tokenizing it and writing the tokens into new symbolic subdivision tier (the example above). Or then taking the tokens, sending those to a morphological analysator, and then writing the result into new tier below. Now pympi has not been used in this, but I would be more than happy to shift into it in this scenario too, as it could be more generic approach than what my project has now. I can send you the script my colleagues are writing, but it isn't yet online, so I can't link to it directly.

This is the kind of structure I use:

So everything has symbolic subdivision on word, lemma, pos and morph tiers, and their references point always to the one above. I guess the main change would be to have the ability to specify how the new annotations are added, following the demands of the stereotype used. Maybe this is already somehow taken into account, it is very possible I don't use pympi correctly!

[nikopartanen]

I just added the script my colleagues and I have been using into this repository, there are also links to few papers where we have described the workflow:

https://github.com/langdoc/elan-fst

So in principle it has two distinct parts, one is reading and writing into ELAN, and another is to send tokens to morphological analysator and parsing the output. We have quite clear plans how to work further with the analysator part, but I'm just thinking that rewriting the ELAN manipulation part with pympi could make it more generalizable and compact. Just adding this information here as it connects to what I was describing earlier in this issue.

[nikopartanen]

Hi! I just would like confirm if this feature works now? I have updated pympi, but I still get the error ValueError: There is no annotation to reference to. when I try to add child annotations at deeper levels. Is there an example somewhere how this can be done correctly? Thanks!

[dopefishh]

Hi, PR #14 didn't solve it apparently I'm afraid. I'll reopen this one. I'm happy to accepts PR's to fix this.