Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
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/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/webdataset/webdataset.py", line 81, in _split_generators
                  first_examples = list(islice(pipeline, self.NUM_EXAMPLES_FOR_FEATURES_INFERENCE))
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 32, in _get_pipeline_from_tar
                  fs: fsspec.AbstractFileSystem = fsspec.filesystem("memory")
                                                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/fsspec/registry.py", line 302, in filesystem
                  cls = get_filesystem_class(protocol)
                File "/usr/local/lib/python3.14/site-packages/fsspec/registry.py", line 239, in get_filesystem_class
                  raise ValueError(f"Protocol not known: {protocol}")
              ValueError: Protocol not known: memory
              
              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 66, 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.

SADAR Analyst Console public evidence release

This card describes the published schema-v4 archive for SADAR Analyst Console, a research demonstrator for post-flight inspection of ADS-B-observable approach criteria at Madrid-Barajas Airport (LEMD).

The archive is a dataset/application evidence bundle, not a Hugging Face model, hosted inference endpoint, or model-serving package. The rules-first application does not load the historical LSTM model.

Release identity

  • Dataset repository: Txemapuch/sadar-analyst-console-release
  • Artifact: sadar-approach-public-release.tar.gz
  • Artifact revision: 820e1715d7e9256c9c9c9cc6d3963629e8cd6c69
  • Release ID: d3474fc76589e43382dc
  • Schema: 4
  • Release kind: sadar_approach_public_evidence
  • Deterministic archive SHA-256: 224801026b450b9ab42ab1bbdb3757086d84061ab73528996b7963bd858e1756
  • Synthetic generator: sadar_synthetic_approach_v2
  • Seed: 20260718
  • OpenSky publication notice: sent 2026-07-20; acknowledgement is not claimed
  • Hub publication status: public, anonymously verified
  • Reviewed source revision: 9fe03e9d144eadfd00fce9a7f8b407698acee30b

The immutable revision was anonymously redownloaded and revalidated before the product lock was replaced.

Three evidence lanes

Lane Included in archive Meaning
Deterministic synthetic demo Yes Fourteen generated scenarios, one attempt/case/operation each, for exercising the analyst workflow. Their position, speed, track, altitude and vertical-rate channels share one runway-relative kinematic construction. They are not recorded flights and do not estimate prevalence.
Aggregate real-data research Yes Suppression-safe counts, rates, coverage, provenance and limitations derived from real OpenSky research cohorts. No row, trajectory, aircraft identifier or exact source timestamp is included.
Ephemeral user upload No Bounded CSV or Parquet data evaluated in memory. Inputs and results are not intentionally retained and never alter the demo queue or aggregate findings.

The eight payload files are demo/catalog.json, demo/attempts.json, demo/cases.json, demo/operations.json, research/aggregate-results.json, config/approach-config.json, config/lemd-geometry.json, and reference/approach-reference.json. A release manifest binds every byte and contract.

Qualification

This candidate is not_qualified_no_independent_labels_or_fresh_holdout. Its only allowed role is research_and_evidence_labeling_demonstrator.

Blocked uses:

  • operational monitoring;
  • emergency detection;
  • stabilized-approach certification;
  • ATC decision support; and
  • safety-performance claims.

There are no independent analyst labels from which to estimate precision, recall, AUROC, calibration or safety performance. The 2026 cohort is already burned, the contextual comparison has no fresh untouched holdout, and status transitions measure rule/reference behavior rather than correctness.

Data source, access and notice

The real aggregate lane was derived from OpenSky Network ADS-B observations around LEMD. To obtain source observations, use OpenSky data access directly and comply with the OpenSky terms of use.

Publication notice status is sent, dated 2026-07-20. This card does not claim that OpenSky has acknowledged the notice.

OpenSky citation:

Matthias Schäfer, Martin Strohmeier, Vincent Lenders, Ivan Martinovic, and Matthias Wilhelm. “Bringing Up OpenSky: A Large-scale ADS-B Sensor Network for Research.” IPSN 2014.

What the application does

  1. Shows deterministic, internally consistent synthetic scenarios in a queue and analyst dossier.
  2. Separates insufficient observation quality from observed criterion evidence.
  3. Uses runway-relative geometry and transparent rules for lateral path, barometric path proxy, observed descent rate, ground-speed envelope and late track correction. The dossier plots those five synchronized signals and marks persistent review intervals on the signal that produced them.
  4. Shows real research findings only as aggregates with denominators, suppression and interpretation limits.
  5. Evaluates a user-supplied bounded file ephemerally against the published rules and aggregate-derived reference parameters.

It does not detect emergencies, establish stabilized-approach compliance, infer intent, clearance, configuration or mass, or certify operational safety.

Withdrawal history

The former schema-3 application archive was withdrawn from the public-delivery path because it contained row-level OpenSky-derived observations, including trajectory and aircraft fields. It must not be mirrored or used as the basis for a new public release. The schema-v2 LSTM demo bundle and Phase-6 training archive remain historical research artifacts only; neither can affect the current Analyst Console verdict or priority.

Reproducibility

Source Git explains and reproduces the release; the dataset registry stores the generated immutable archive. A clean checkout deterministically generates the fourteen synthetic cases without reading row-level source data, checks cross-channel kinematic consistency, combines them with the tracked aggregate resource, validates schema 4, rebuilds the archive and proves its digest matches the public lock. Production uses the explicit locked-public path and fails closed if that immutable schema-v4 lock drifts.

Downloads last month
69