texttheater / ccgweb Goto Github PK
View Code? Open in Web Editor NEWA Web platform for parsing and annotation
A Web platform for parsing and annotation
https://texttheater.net/ccgweb/sentence.php?lang=eng&sentence=A%20glass%20of%20water%2C%20please.
What final category these phrases should have?
S[b] or S[b]\NP OR NP?
If this phrase is uttered without any previous discourse, I can understand it as imperative.
But if it is an reply to "What would you like to drink?", then it is more NP.
https://texttheater.net/ccgweb/sentence.php?lang=deu&sentence=Sie%20kommen%20nicht%20heute.
Is it ok if nicht takes scope over heute?
Otherwise VP/VP for 'nicht' seems weird, right?
I annotated noch eine
as a constituent but maybe I shouldn’t.
In CCGrebank, wh-questions have category S[wq], with an intermediary S[q] layer.
Should Italian use S[dcl] instead of S[q]?
Should Italian yes/no question with declarative word order use S[dcl] instead of S[q]?
Doesn't do desired forward cross composition over plu usato and nel_mondo
Just pasting the URL leads to GitHub cutting off sentence-final puncutation.
Judge says 2 arguments
https://texttheater.net/ccgweb/sentence.php?lang=eng&sentence=I%20will%20take%20care%20of%20the%20flowers.
In the manual it is analyzed with 3 arguments:
https://texttheater.net/ccgweb/manual.php?section=arguments-and-modifiers
This sentence and its translations: the embedded sentences are strictly speaking missing an S[adj]\NP
argument. Should that be accounted for in the categories?
Also, is als ze er uitziet an argument of zo (similar for the other languages)?
CCGrebank says no, my linguistic ethos says yes.
A new rule might be needed for handling predicative adverbs and adjectives with omitted copula, which is very frequent in Russian.
Currently BOWs are not visible and probably it is not important that they will be.
Thought It will be nice that one can undo all its edits and get the derivation on its tab that is the same as what is on the auto tab.
One can get auto tree on its tab by copying all (in the worst case) the lexical categories and spans from the auto, but that can be very unpleasant task when sentences are long.
How would you get "sollen" with a wide scope?
or are we accepting the derivation that leads the following composition?
(nicht (hätte @ (sollen @ (ausklinken @ mich)))) @ Ich
Is this the desired analysis?
N/PP N\(N/PP)
------ --------
Plätze Tokyos
Problem: EasyCCG thinks the second category is type-raised and therefore doesn’t allow backward application.
Tokens should run from right to left and slashes should be flipped accordingly, I think. But not sure about functional categories yet: should functors remain on the left or move to the right?
Unnecessary type raising on "Io".
Should re-enable creation of .raw
files by the REST server.
https://texttheater.net/ccgweb/sentence.php?lang=eng&sentence=Don%27t%20fuck%20it%20up%21
Cannot choose the German translation "Versau es nicht!" as I put ")" as a category:
ERROR: bad API response status.
I was wondering whether . is the umbrella category for all punctuations.
Cause in rebanked CCG comma, period, quotes and brackets have different categories.
Entering CONJ doesn't work, need to enter conj.
Some kind of indeterminism in this example:
https://texttheater.net/ccgweb/sentence.php?lang=eng&sentence=It%20was%20colder%20yesterday%20than%20today.
(1) When I constraint the whole sentence as a constituent "colder yesterday than today" stays as a constituent.
(2) When "it"'s category is set to NP[expl], "was colder yesterday" becomes a constituent.
(3) When I set the category back to NP, the "was colder yesterday" still stays as a constituent, opposed to what was after (1).
Non-canonical argument order. How to analyze this?
zum gets PP/N
, taxi N
, un NP/N
and there is no parse.
Set nehmen to ((S[b]/NP)/PP)/NP
and Sie to NP
. Still no parse.
Set Hotel to N
and . to S[b]\S[b]
. Still no parse.
The reason can be that there is no type-raising rule: NP --> S\(S/NP)
"ich mache" should combine via forward type-raising and forward composition, not via backward application. Currently there is no way to tell the system this.
We need a rule here: is di always a preposition, always a complementizer, or does it depend?
https://texttheater.net/ccgweb/sentence.php?lang=eng&sentence=Does%20she%20speak%20English%3F
Why not
Does_S[q]/S[b] (she speak English)_S[b]
instead of
(Does_(S[q]/(S[b]\NP))/NP she) (speak English)_S[b]\NP
Cf. #28 - would be nice if annotators could introduce such rules themselves, and make sure the parser actually picks them.
Currently the download only includes BOWs, not full analyses.
No indentations and newlines, currently. Maybe use SWI-Prolog’s SGML library?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.