Functions to analyse and manipulate dependency trees, as well as conversions between GF and dependency trees. The main use case is UD (Universal Dependencies), but the code is designed to be completely generic as for annotation scheme. This repository replaces the old gf-contrib/ud2gf code. It is also meant to be used in the 'vd' command of GF and replace the supporting code in gf-core in the future.
We have this standard way of distinguishing between singular and plural the:
#auxcat The DET
#auxfun DetCN_theSg det cn : The -> CN -> NP = DetCN the_Det cn ; det head[Number=Sing]
#auxfun DetCN_thePl det cn : The -> CN -> NP = DetCN thePl_Det cn ; det head[Number=Plur]
#disable the_Det thePl_Det
Now I would like to do the same for other determiners that are ambiguous for number, like some and any. My file is this:
1 any any DET DT _ 2 det _ _
2 word word NOUN NN Number=Sing 0 root _ _
1 any any DET DT _ 2 det _ _
2 words word NOUN NN Number=Plur 0 root _ _
1 some some DET DT _ 2 det _ _
2 word word NOUN NN Number=Sing 0 root _ _
1 some some DET DT _ 2 det _ _
2 words word NOUN NN Number=Plur 0 root _ _
The naive way would be to do something like this. (The is still the auxcat for DET from the previous example.)
#auxfun DetCN_anySg det cn : The -> CN -> NP = DetCN anySg_Det cn ; det head[Number=Sing]
#auxfun DetCN_anyPl det cn : The -> CN -> NP = DetCN anyPl_Det cn ; det head[Number=Plur]
#disable anyPl_Det anySg_Det
#auxfun DetCN_someSg det cn : The -> CN -> NP = DetCN someSg_Det cn ; det head[Number=Sing]
#auxfun DetCN_somePl det cn : The -> CN -> NP = DetCN somePl_Det cn ; det head[Number=Plur]
#disable somePl_Det someSg_Det
With this, I get the following results—the auxfuns didn't seem to do anything. Looking at dt and bt, I see no auxfuns being used.
DetCN anyPl_Det (UseN word_N)
LIN: any words
DetCN anyPl_Det (UseN word_N)
LIN: any words
DetCN somePl_Det (UseN word_N)
LIN: some words
DetCN somePl_Det (UseN word_N)
LIN: some words
So I try to change the auxfuns into this: taking as an argument an actual RGL cat Det, not auxcat The (which corresponds to a DET).
#auxfun DetCN_anySg det cn : Det -> CN -> NP = DetCN anySg_Det cn ; det head[Number=Sing]
#auxfun DetCN_anyPl det cn : Det -> CN -> NP = DetCN anyPl_Det cn ; det head[Number=Plur]
#disable anyPl_Det anySg_Det
#auxfun DetCN_someSg det cn : Det -> CN -> NP = DetCN someSg_Det cn ; det head[Number=Sing]
#auxfun DetCN_somePl det cn : Det -> CN -> NP = DetCN somePl_Det cn ; det head[Number=Plur]
#disable somePl_Det someSg_Det
Now I see, from looking at bt0, that the auxfuns take action:
bt0: DetCN_anySg anyPl_Det (UseN word_N)
at: DetCN anySg_Det (UseN word_N)
LIN: any word
bt0: DetCN_anyPl anyPl_Det (UseN word_N)
at: DetCN anyPl_Det (UseN word_N)
LIN: any words
But unfortunately, there is no matching with strings, so the auxfun DetCN_any* auxfuns take action even when the actual Det is some*_Det.
bt0: DetCN_anySg somePl_Det (UseN word_N)
at: DetCN anySg_Det (UseN word_N)
LIN: any word
bt0: DetCN_anyPl somePl_Det (UseN word_N)
at: DetCN anyPl_Det (UseN word_N)
LIN: any words
So I would like to enhance the macro DSL such that we can add a wordform or lemma constraint among the tag constraints. For example (feel free to suggest a better syntax)
#auxfun DetCN_anySg det cn : Det -> CN -> NP = DetCN anySg_Det cn ; det head[Number=Sing|wf="any"]
#auxfun DetCN_anyPl det cn : Det -> CN -> NP = DetCN anyPl_Det cn ; det head[Number=Plur|wf="any"]
#disable anyPl_Det anySg_Det
#auxfun DetCN_someSg det cn : Det -> CN -> NP = DetCN someSg_Det cn ; det head[Number=Sing|wf="some"]
#auxfun DetCN_somePl det cn : Det -> CN -> NP = DetCN somePl_Det cn ; det head[Number=Plur|wf="some"]
#disable somePl_Det someSg_Det
We have this standard way of distinguishing between singular and plural the:
Now I would like to do the same for other determiners that are ambiguous for number, like some and any. My file is this:
The naive way would be to do something like this. (
Theis still the auxcat for DET from the previous example.)With this, I get the following results—the auxfuns didn't seem to do anything. Looking at
dtandbt, I see no auxfuns being used.So I try to change the auxfuns into this: taking as an argument an actual RGL cat
Det, not auxcatThe(which corresponds to a DET).Now I see, from looking at
bt0, that the auxfuns take action:But unfortunately, there is no matching with strings, so the auxfun
DetCN_any*auxfuns take action even when the actual Det issome*_Det.So I would like to enhance the macro DSL such that we can add a wordform or lemma constraint among the tag constraints. For example (feel free to suggest a better syntax)