The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 291, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Column() changed from object to string in row 0
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 101, in _split_generators
pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 306, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.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.
The Stochastic Parrot — cross-outlet discrepancy corpus
An automated media-criticism desk reads how multiple newsrooms cover the same event and logs every place their accounts diverge, anchored to verbatim quotes from frozen snapshots of each source page.
The headline finding: most press disagreement is about the word, not the event. Across the corpus, 92% of contested points are framing or naming splits; only 7.9% are hard contradictions where outlets assert incompatible facts.
As of 2026-11-01:
| Measure | Count |
|---|---|
| Pieces with a cross-outlet discrepancy matrix | 1,139 |
| Contested points (grouped splits) | 3,653 |
| Hard contradictions (outlets assert incompatible facts) | 287 |
| Naming splits (one referent, two labels) | 1,151 |
| Framing splits (same facts, different word) | 2,215 |
- Framing rate: 92%
- Newsrooms read: 265
- Corrections the desk has published about its own work: 49
Files
| File | What it is |
|---|---|
corpus.json |
The full structured record — every piece, every split, every source span. |
framing-index.json |
The longitudinal measurement: totals, a monthly trend, and a per-outlet table. |
framing-index-by-outlet.csv |
The per-outlet table as tidy CSV (outlet, hard, framing, total). |
lexicon.json |
The dictionary of contested language: every term the press has used to mean two things. |
flags.json |
Flagged words by kind (euphemism, characterization-as-fact, ambiguity). |
angles.json |
The desk's angle ledger. |
corrections.json |
Every correction published, dated, with what was wrong and what changed. |
How the taxonomy works
A hard contradiction means two outlets assert incompatible facts about the same thing; it is the only case the desk calls a contradiction. A framing split means outlets agree on the facts and differ in the word they chose. A naming split means one referent carries two labels, each valid at a different level of description. Every quoted span is verbatim and carries a character offset into a frozen snapshot, so any claim here is independently checkable.
Licence and citation
CC BY 4.0. Cite as: The Stochastic Parrot, cross-outlet discrepancy corpus, 2026-11-01, https://thestochasticparrot.com. Method: https://thestochasticparrot.com/method · Corrections: https://thestochasticparrot.com/corrections/ · Index: https://thestochasticparrot.com/framing-index/
Regenerated from the live site by the desk's own build; this repo is a mirror, never the source.
- Downloads last month
- 164