Hugging Face's logo

Datasets:
thodsapon
/
thai-ncb-digital-fraud-structural-rag

Tasks:
Text Retrieval
Languages:
Thai
Size:
n<1K
Tags:
thai
legal
regulatory
rag
education
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
schema_version: int64
corpus_scope: string
source_filename: string
source_file_hash: string
issuer: string
law_name: string
source_url: string
source_page_count: int64
included_pages: list<item: int64>
  child 0, item: int64
excluded_pages: string
extraction_method: string
parent_provisions: int64
retrieval_children_x_x_only: int64
deep_child_policy: string
parent_child_contract: string
embedding_model: string
embedding_endpoint: string
embedding_dimensions: int64
embedding_normalization: string
raw_embedding_norm_min: double
raw_embedding_norm_max: double
created_at: string
embedding_provenance: string
parent_id: string
heading: string
node_role: string
citation_label: string
node_type: string
parent_ref: struct<id: string, node_type: string, structural_id: string, heading: string, structural_path: list< (... 14 chars omitted)
  child 0, id: string
  child 1, node_type: string
  child 2, structural_id: string
  child 3, heading: string
  child 4, structural_path: list<item: string>
      child 0, item: string
retrieval_allowed: bool
content: string
content_sha256: string
citation_id: string
source_sha256: string
structural_path: list<item: string>
  child 0, item: string
record_id: string
dataset_id: string
source_section_ids: list<item: string>
  child 0, item: string
page_start: int64
page_end: int64
passage_text: string
topic: string
to
{'dataset_id': Value('string'), 'record_id': Value('string'), 'node_role': Value('string'), 'node_type': Value('string'), 'retrieval_allowed': Value('bool'), 'citation_label': Value('string'), 'citation_id': Value('string'), 'source_section_ids': List(Value('string')), 'law_name': Value('string'), 'heading': Value('string'), 'topic': Value('string'), 'parent_id': Value('string'), 'parent_ref': {'id': Value('string'), 'node_type': Value('string'), 'structural_id': Value('string'), 'heading': Value('string'), 'structural_path': List(Value('string'))}, 'structural_path': List(Value('string')), 'page_start': Value('int64'), 'page_end': Value('int64'), 'source_url': Value('string'), 'source_sha256': Value('string'), 'content_sha256': Value('string'), 'content': Value('string'), 'passage_text': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 489, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2818, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              schema_version: int64
              corpus_scope: string
              source_filename: string
              source_file_hash: string
              issuer: string
              law_name: string
              source_url: string
              source_page_count: int64
              included_pages: list<item: int64>
                child 0, item: int64
              excluded_pages: string
              extraction_method: string
              parent_provisions: int64
              retrieval_children_x_x_only: int64
              deep_child_policy: string
              parent_child_contract: string
              embedding_model: string
              embedding_endpoint: string
              embedding_dimensions: int64
              embedding_normalization: string
              raw_embedding_norm_min: double
              raw_embedding_norm_max: double
              created_at: string
              embedding_provenance: string
              parent_id: string
              heading: string
              node_role: string
              citation_label: string
              node_type: string
              parent_ref: struct<id: string, node_type: string, structural_id: string, heading: string, structural_path: list< (... 14 chars omitted)
                child 0, id: string
                child 1, node_type: string
                child 2, structural_id: string
                child 3, heading: string
                child 4, structural_path: list<item: string>
                    child 0, item: string
              retrieval_allowed: bool
              content: string
              content_sha256: string
              citation_id: string
              source_sha256: string
              structural_path: list<item: string>
                child 0, item: string
              record_id: string
              dataset_id: string
              source_section_ids: list<item: string>
                child 0, item: string
              page_start: int64
              page_end: int64
              passage_text: string
              topic: string
              to
              {'dataset_id': Value('string'), 'record_id': Value('string'), 'node_role': Value('string'), 'node_type': Value('string'), 'retrieval_allowed': Value('bool'), 'citation_label': Value('string'), 'citation_id': Value('string'), 'source_section_ids': List(Value('string')), 'law_name': Value('string'), 'heading': Value('string'), 'topic': Value('string'), 'parent_id': Value('string'), 'parent_ref': {'id': Value('string'), 'node_type': Value('string'), 'structural_id': Value('string'), 'heading': Value('string'), 'structural_path': List(Value('string'))}, 'structural_path': List(Value('string')), 'page_start': Value('int64'), 'page_end': Value('int64'), 'source_url': Value('string'), 'source_sha256': Value('string'), 'content_sha256': Value('string'), 'content': Value('string'), 'passage_text': Value('string')}
              because column names don't match

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.

Thai NCB and BOT Digital Fraud Structural RAG Corpus

Two small, page-anchored Thai corpora prepared for retrieval experiments and classroom-style RAG work. The aim is to make the source structure easier to explore while keeping a clear path back to the original document.

About the experiment

This dataset sits behind a small Qwen3.6-35B-A3B Q5 versus OpenThai 2.0 Legal BF16 test. The short write-up, result table and UI captures are on GitHub: https://github.com/lengtsp/OpenThai-2.0-Legal-Test-Result

This material is shared for education and experiment only. The original legal and notice text is not rewritten or substantively changed here; the dataset only arranges it into a traceable parent-and-child structure for retrieval. The original BOT documents remain the authoritative source.

Who put this together

This dataset and the linked GitHub page were put together by an independent tester as part of the experiment. They are not official model repositories, and the tester is not part of, employed by, or speaking for the teams that develop or publish Qwen or OpenThai 2.0 Legal on Hugging Face or GitHub.

Contents

Config Records Indexable children Structure
ncb_credit_info_act_2559 76 66 9 หมวด parents + one preliminary group; primary มาตรา children only
bot_digital_fraud_management_2568 17 11 6 top-level parents; ข้อ X.X children only

Only rows where retrieval_allowed=true may be embedded/indexed. Parent rows have empty content and passage_text, preventing duplicate parent/child embeddings.

How the structure works

  • NCB: official BOT copy, pp. 1–18. Exactly 66 primary article children. Source articles 20/1 and 31/1 remain in their adjacent primary groups (20 and 31); headers/footers are excluded.
  • Digital Fraud: BOT Notice 57/2568, pp. 2–13. Exactly 11 retrieval/citation children at level X.X. X.X.X and lower material remains inside the content of its owning X.X child and is never an independent row.
  • Every child carries parent_id and parent_ref for direct, inspectable provenance.

Sources

Each record points back to its source URL, page anchor and source SHA-256. The original PDFs are not bundled here. The card does not change the rights attached to the source documents.

Data fields

dataset_id, record_id, node_role, node_type, retrieval_allowed, citation_label, citation_id, source_section_ids, law_name, heading, topic, parent_id, parent_ref, structural_path, page_start, page_end, source_url, source_sha256, content_sha256, content, and passage_text.

No question, answer, reference answer, expected citation, model output, embedding vector, or benchmark score is included.
Downloads last month
25
Total file size:
557 kB