search

Annotald / annotald

A program for annotation in the Penn Treebank format
GNU General Public License v3.0
8 stars 3 forks source link

Question about hash-script #73

Open diertani opened 9 years ago

diertani commented 9 years ago

This is not a problem, just a question from sheer curiosity. The script that alerts me to changes in the tokens doesn't alert me to changes if I split a sentence token from the front, but does if I do from the back. What's the difference, from the perspective of the script?

aecay commented 9 years ago

It's not supposed to alert you in either case; token split/join operations are not supposed to affect it. So this is a problem. What is the difference between splitting "from the front" vs. "from the back"?

diertani commented 9 years ago

Ah! Well, in that case I'm glad I asked!

The difference is: if I have

(XP (YP) (ZP))

and I remove YP into a new root token, I don't get the message the next time I save, but if I move out ZP, I do. I also get it if I demote a root-level token to the status of a consituent in another token, from either direction, and if I combine two tokens into a single token, but not if I combine that new token with another token.

aecay commented 9 years ago

That's very odd...the result is the same in both cases isn't it:

(XP (YP))
(WP (ZP))

I'll have to look into and fix this...

diertani commented 9 years ago

Also, it seems to be the case that once I've experimentally combined tokens, it gives me the error message when I remove constituents from either direction.

diertani commented 9 years ago

Yeah, the result is the same...

diertani commented 9 years ago

...and also if I delete a spurious XP created through experimentation, it gives me the error message even though in that case the text has not changed at all.