Approach for tagging grammatical patterns (Japanese)

[e-e]

I'm looking for some direction regarding the best approach to defining and tagging grammatical patterns in Japanese. I have taken a bit of a break since the last time I was toying with spaCy for this task, so I'm a bit rusty, but I know previously I was wondering if Prodigy would be a good approach for this. I also attempted manually constructing matchers, but wasn't totally able to figure out some edge cases (very likely a result of not being super familiar with the library though). Anyways, any help regarding an approach would be very much appreciated - thank you in advance!

[polm]

What is your actual question / goal? I don't remember what you are referring to. The DependencyMatcher should be able to match many types of grammatical patterns.

[e-e]

Essentially I want to tag each grammar point with some unique id which is associated to the grammar point in a database.

[polm]

Since any textbook's list of "grammar points" is to some degree arbitrary, I don't think there's any way you can do better than manually writing rules for them, though it looks like the actual rules should be pretty simple.

For IDs, if you make a list of IDs for each grammar point, you should be able to use the EntityRuler ID feature to map matches back to grammar points.

[e-e]

Would prodigy be useful for this?

[polm]

Prodigy would be useful for reviewing your patterns and checking the results. I guess it depends on what you consider a "grammar point", but I would think a rule-based approach would capture them well and it'd be hard to use a statistical model, so you wouldn't benefit much from training data / a model.

[e-e]

I'll give the rule-based matching a shot. Thanks for the help!