README.md

metadata

language:
  - vi
  - en
license: cc-by-nc-4.0
task_categories:
  - text-classification
tags:
  - esg
  - greenwashing-detection
  - banking
  - vietnam
  - sustainability
size_categories:
  - 1K<n<10K

Vietnamese Banking ESG Disclosure Dataset

Description

Multi-task annotated dataset for ESG (Environmental, Social, Governance) disclosure analysis in the Vietnamese banking sector. Each sample is a text chunk from bank sustainability/annual reports, annotated with:

• Greenwashing classification (Legitimate / Greenwashing / Uncertain)
• ESG pillar (Environmental / Social / Governance / General)
• Content quality score (0-100)
• ESG score (0-100)

Dataset Statistics

Split	Samples
train	2,059
val	558
test	900
Total	3,517

Schema

Field	Type	Description
text	string	ESG disclosure text chunk
greenwash_label	string	Legitimate / Greenwashing / Uncertain
pillar_label	string	Environmental / Social / Governance / General
quality_score	float	Content quality score (0-100)
esg_score	float	Predicted ESG performance score (0-100)
report_type	string	SUSTAINABILITY or ANNUAL
report_year	int	Report fiscal year
bank_id	string	Anonymized bank identifier
assurance_level	string	Third-party assurance level
has_scope3	bool	Contains Scope 3 emission data
gri_compliant	bool	GRI Standards compliant
tcfd_elements_count	int	TCFD recommendation coverage (0-11)
circular17_compliant	bool	SBV Circular 17 compliant

Annotation Process

1. Initial labeling via LLM (Step 2) with rule-based fallback
2. Content quality scoring (Step 3)
3. Human review and validation (Step 4)

Citation

If you use this dataset, please cite:

[Citation to be added upon publication]