cu-clear / semlink Goto Github PK

Hi,

Thanks for releasing this resource! I had a few questions about the PropBank-VerbNet mappings:

1. What might be the reason that a PropBank role's mapping to a VerbNet role is missing? Is it because the corresponding role does not exist in VerbNet or because the annotation hasn't been done yet?
2. It appears that the ARGM-* roles are not mapped to any VerbNet roles. Is there a reason for this?
3. What is the difference between the VerbNet mappings in the frame files here and SemLink?
4. Is there a README describing the format of the latest release of SemLink?

When comparing pb-vn2.json to a mapping roleset-class mapping derived from VerbNet3.4 itself, I find that the ProbBank rolesets in the domain of each mapping are not in a subset relation with each other as might be expected.

To derive the mapping from VerbNet3.4, I use:

from collections import defaultdict
from verbnet import VerbNetParser

verbnet = VerbNetParser(version="3.4")
     
pb_vn34_map = defaultdict(set)

for cid, clsinfo in verbnet.verb_classes_numerical_dict.items():
    for m in clsinfo.members:
        for pbroleset in m.grouping:
            pb_vn34_map[pbroleset] |= {cid}

pb_vn34_map = dict(pb_vn34_map)

When compared to pb-vn2.json...

with open('semlink/instances/pb-vn2.json') as f:
    semlink_map = json.load(f)
    
pbset_from_verbnet = set(pb_vn34_map)
pbset_from_semlink = set(semlink_map)
    
print('In both SemLink and VerbNet:\t', len(pbset_from_semlink & pbset_from_verbnet))
print('In VerbNet but not SemLink:\t', len(pbset_from_verbnet - pbset_from_semlink))
print('In SemLink but not VerbNet:\t', len(pbset_from_semlink - pbset_from_verbnet))

I observe the following counts:

In both SemLink and VerbNet:	 1854
In VerbNet but not SemLink:	 1360
In SemLink but not VerbNet:	 2323

Hi, I had a few more questions about SemLink-2.

1. Why are there much fewer entries in SemLink2 pb-vn mappings? It seems like that a large number of these entries that existed in SemLink 1.2.2 are just missing in SemLink-2. More than 3900 entries are just mapped to empty lists. For eg, "force", "fortify" and many others are mapped to empty lists. Is this intentional?

2. Some verb lemmas have different PropBank rolesets and the same VN-class but the argument mappings only exist for one of these cases in SemLink 1.2.2. Why is this the case? Here is an example:

3. For any given sentence, would it reasonable to map PropBank arguments to VerbNet arguments using the pb-vn type map in SemLink (after having retrieved the correct verb sense)?

Thanks for your help!

Hello,
thanks for this data! I'm trying to map propbank rolesets andarguments to FrameNet frames and frame-elements. But as I see this is not (yet) possible using the .json files in instances/. So I try to use the instances/semlink-2 file (together with tools/annotation.py. However, I came across cases where the FrameNet frame is None (OK, no FrameNet frame defined) NF or IN. What is the meaning of the latter two?

BTW, the python lines indicating how to use the json-files seem to have an error, the following works:

    # Let's get the FN mappings from the PB roleset "abduct.01"
    vn_mapping = pb_vn_mappings["abduct.01"]    

    # Here we just grab the first sense
    # vn_class = vn_mapping[list(vn_mapping.keys())[0]] # original line (here vn_class is a dict)
    vn_class = list(vn_mapping.keys())[0]  # should be like this

    # From VN, abduct.01 maps to VerbNet class 10.5
    fn_mapping = vn_fn_mappings[vn_class + "-" + verb]

    print (fn_mapping)
    # And from FN we get the Frame "kidnapping"

Thanks

Hi, in https://verbs.colorado.edu/semlink/, we have WN listed. But I didn't see the links to WN in this repo. Can you clarify? WN was linked? What version?

Dear Mrs. Kazeminejad, dear Mr. Stowe,

I am a PhD student in computational linguistics from Germany. I needed mappings from WordNet to VerbNet for my project, so I scraped them with a web crawler from this page: https://verbs.colorado.edu/html_groupings/ . The mappings can be found in my repository WordNet2VerbNet_Mappings .

I would be glad if you could find the mappings useful. If not, please feel free to delete my comment.

Sincerely yours,
Tatiana Bladier

Hi,

This is a great resource, but looking through the code in the tools folder, they are full of absolute paths on someone's computer and refer to resources that do not seem to exist here.

There seems to be no indication at all of how you can use the mappings files. Looking at them doesn't provide a lot of insight. It's pretty hard to use if there is no information.

I checked docs for older versions, but they refer to files that don't seem to exist anymore.

Does anyone know of even just a simple example of using the mappings between Framenet and Propbank (including roles) that exists anywhere and that matches with this version?

Thanks!
Alan

In March, I pointed out to @ghamzak in an email discussion that the PB-VN mapping were treated as one-to-one but that they must be one-to-many because the same PB roleset can map to distinct VN classes. This commit fixed the problem, but the error was reintroduced by this commit (@kevincstowe). I might suggest developing a JSON schema against which these mappings are validated prior to merging into master to avoid reintroducing this error in the future.

Will it be possible to get the full text annotated data. The annotation file does not contain full text

Hi,

I'm trying to use SemLink data to map a FrameNet frame with its members to a set of PropBank predicates and roles.
Currently, in version 2.0 in GitHub, I don't see any specific role mappings between the two, only in SemLink 1.2.2c I see some.
Any chance to include them in GitHub too?
Also - an official example script of how to get such a mapping would be really useful to get onboarded.

Cheers!