Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed because of a cast error
Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 3 new columns ({'parent', 'child', 'label'}) and 2 missing columns ({'parent_synset_uri', 'child_synset_uri'}).

This happened while the csv dataset builder was generating data using

hf://datasets/loraug/wordnet-neural/finetuning/oewn_hypernymy_negatives/oewn_train_negatives_seed_42.csv (at revision 2d0242118e9b1e649902e446ea2f6b1e2631b39c), ['hf://datasets/loraug/wordnet-neural@2d0242118e9b1e649902e446ea2f6b1e2631b39c/finetuning/oewn_hypernymy_datasets/oewn_hypernymy_train.csv', 'hf://datasets/loraug/wordnet-neural@2d0242118e9b1e649902e446ea2f6b1e2631b39c/finetuning/oewn_hypernymy_negatives/oewn_train_negatives_seed_42.csv']

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1848, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 784, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 795, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              child: string
              parent: string
              label: int64
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 610
              to
              {'child_synset_uri': Value('string'), 'parent_synset_uri': Value('string')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1850, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
                  ...<4 lines>...
                  )
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 3 new columns ({'parent', 'child', 'label'}) and 2 missing columns ({'parent_synset_uri', 'child_synset_uri'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/loraug/wordnet-neural/finetuning/oewn_hypernymy_negatives/oewn_train_negatives_seed_42.csv (at revision 2d0242118e9b1e649902e446ea2f6b1e2631b39c), ['hf://datasets/loraug/wordnet-neural@2d0242118e9b1e649902e446ea2f6b1e2631b39c/finetuning/oewn_hypernymy_datasets/oewn_hypernymy_train.csv', 'hf://datasets/loraug/wordnet-neural@2d0242118e9b1e649902e446ea2f6b1e2631b39c/finetuning/oewn_hypernymy_negatives/oewn_train_negatives_seed_42.csv']
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

child_synset_uri
string
parent_synset_uri
string
https://en-word.net/id/oewn-00001930-n
https://en-word.net/id/oewn-00001740-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00001740-n
https://en-word.net/id/oewn-00002452-n
https://en-word.net/id/oewn-00001930-n
https://en-word.net/id/oewn-00002684-n
https://en-word.net/id/oewn-00001930-n
https://en-word.net/id/oewn-00003993-n
https://en-word.net/id/oewn-00003553-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00004258-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00007347-n
https://en-word.net/id/oewn-00005787-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00005930-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00006024-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00006150-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00006269-n
https://en-word.net/id/oewn-00004258-n
https://en-word.net/id/oewn-00006400-n
https://en-word.net/id/oewn-00004258-n
https://en-word.net/id/oewn-00006484-n
https://en-word.net/id/oewn-00004258-n
https://en-word.net/id/oewn-00007347-n
https://en-word.net/id/oewn-00001930-n
https://en-word.net/id/oewn-00007846-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00015568-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00017402-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00019226-n
https://en-word.net/id/oewn-00004475-n
https://en-word.net/id/oewn-00019308-n
https://en-word.net/id/oewn-00003553-n
https://en-word.net/id/oewn-00019793-n
https://en-word.net/id/oewn-00021007-n
https://en-word.net/id/oewn-00019793-n
https://en-word.net/id/oewn-13831419-n
https://en-word.net/id/oewn-00020270-n
https://en-word.net/id/oewn-00021007-n
https://en-word.net/id/oewn-00021007-n
https://en-word.net/id/oewn-00001930-n
https://en-word.net/id/oewn-00021445-n
https://en-word.net/id/oewn-00020270-n
https://en-word.net/id/oewn-00021914-n
https://en-word.net/id/oewn-00020270-n
https://en-word.net/id/oewn-00022119-n
https://en-word.net/id/oewn-00003553-n
https://en-word.net/id/oewn-00023083-n
https://en-word.net/id/oewn-00022119-n
https://en-word.net/id/oewn-00023280-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00023451-n
https://en-word.net/id/oewn-00023280-n
https://en-word.net/id/oewn-00023953-n
https://en-word.net/id/oewn-00023280-n
https://en-word.net/id/oewn-00024444-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00026390-n
https://en-word.net/id/oewn-00024900-n
https://en-word.net/id/oewn-00027365-n
https://en-word.net/id/oewn-00002684-n
https://en-word.net/id/oewn-00028468-n
https://en-word.net/id/oewn-00024444-n
https://en-word.net/id/oewn-00028950-n
https://en-word.net/id/oewn-00024444-n
https://en-word.net/id/oewn-00029306-n
https://en-word.net/id/oewn-00028950-n
https://en-word.net/id/oewn-00029677-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00029677-n
https://en-word.net/id/oewn-00031563-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00032220-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00032912-n
https://en-word.net/id/oewn-00032220-n
https://en-word.net/id/oewn-00033122-n
https://en-word.net/id/oewn-00032220-n
https://en-word.net/id/oewn-00033319-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00033914-n
https://en-word.net/id/oewn-00002137-n
https://en-word.net/id/oewn-00034512-n
https://en-word.net/id/oewn-00029976-n
https://en-word.net/id/oewn-00034778-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00034873-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00035041-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00035244-n
https://en-word.net/id/oewn-00408356-n
https://en-word.net/id/oewn-00035660-n
https://en-word.net/id/oewn-00210792-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00036418-n
https://en-word.net/id/oewn-00428985-n
https://en-word.net/id/oewn-00036612-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00036782-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00037020-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00037301-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00037483-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00037727-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00037811-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00037921-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00038913-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00039000-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00039311-n
https://en-word.net/id/oewn-00039000-n
https://en-word.net/id/oewn-00039759-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00039949-n
https://en-word.net/id/oewn-00039759-n
https://en-word.net/id/oewn-00040035-n
https://en-word.net/id/oewn-00039759-n
https://en-word.net/id/oewn-00040283-n
https://en-word.net/id/oewn-00040035-n
https://en-word.net/id/oewn-00040478-n
https://en-word.net/id/oewn-00040035-n
https://en-word.net/id/oewn-00040654-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00040728-n
https://en-word.net/id/oewn-00040035-n
https://en-word.net/id/oewn-00040890-n
https://en-word.net/id/oewn-00039759-n
https://en-word.net/id/oewn-00041158-n
https://en-word.net/id/oewn-00040890-n
https://en-word.net/id/oewn-00041542-n
https://en-word.net/id/oewn-00040890-n
https://en-word.net/id/oewn-00041700-n
https://en-word.net/id/oewn-00041542-n
https://en-word.net/id/oewn-00042206-n
https://en-word.net/id/oewn-00038116-n
https://en-word.net/id/oewn-00042352-n
https://en-word.net/id/oewn-00743943-n
https://en-word.net/id/oewn-00042637-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00043279-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00043495-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00043854-n
https://en-word.net/id/oewn-00037483-n
https://en-word.net/id/oewn-00043933-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00044347-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00044640-n
https://en-word.net/id/oewn-00037483-n
https://en-word.net/id/oewn-00044888-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00045193-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00045411-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00045855-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00045991-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00046387-n
https://en-word.net/id/oewn-00037483-n
https://en-word.net/id/oewn-00046648-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00046918-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00047085-n
https://en-word.net/id/oewn-00037483-n
https://en-word.net/id/oewn-00047263-n
https://en-word.net/id/oewn-00030657-n
https://en-word.net/id/oewn-00047759-n
https://en-word.net/id/oewn-00037483-n
https://en-word.net/id/oewn-00047847-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00048097-n
https://en-word.net/id/oewn-00047847-n
https://en-word.net/id/oewn-00048682-n
https://en-word.net/id/oewn-00035910-n
https://en-word.net/id/oewn-00048792-n
https://en-word.net/id/oewn-00035910-n
End of preview.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

How Neural is WordNet?

A Comparison between Human Hierarchical Semantic Representations and Pretrained Language Models Embeddings

Pretrained language models encode rich semantic information, but it is unclear whether their representations capture the hierarchical organization of the lexicon or only reflect semantic similarity.

We propose to investigate this with the experiments documented in this repository.

We compare the Open English WordNet (https://en-word.net/) noun hierarchy with embeddings derived from BERT-base-uncased (https://huggingface.co/google-bert/bert-base-uncased).


The data folder contains a Turtle file with information about the linking between Open English WordNet synsets and their vectorial representations derived from BERT-base-uncased. The synsets are organized through the hypernymy relation as encoded in OEWN, and to each pair a cosine similarity score is attached.

The finetuning folder contains two subdirectories:

  • oewn_hypernymy_datasets contains the training, development and tes set, composed of respectively 59,419, 7,400 and 7,404 edges.
  • oewn_hypernymy_negatives which contains the negative samples for each of the splits.

A collection of notebooks used in the experiments is available in the notebooks folder:

  • bert_noun-def_embeddings is used to derive bert embeddings for every wordnet noun-definition pair.
  • evaluate_hypernymy_multiple_seeds is used to test three different configurations in the task of hypernymy discovery.
  • finetune_bert_oewn_hypernymy is used to fine-tune BERT on noun hypernymy and test it in the task of hypernymy discovery, before and after training.
  • rdf2vec_transe_geometry is used to derive vectorial representations of the wordnet hierarchy using two different methods, namely RDF2Vec and TransE.
  • oewn_hypernym_relation_correlation is used to compute correlation metrics between the different vectorial representations of the noun hierarchy previously obtained from BERT, RDF2Vec and TransE.
  • hypernymy_graph_compatible is used to evaluate seven configurations in a graph compatible hypernymy detection task.
Downloads last month
45