IndoBias contains stereotypical statements about Indonesian social groups that may be offensive. Access is granted for research on measuring and mitigating bias. Requests are reviewed manually by the authors.

By requesting access you agree to use IndoBias only for research on evaluating or mitigating social bias, not to train models to produce stereotypes, not to redistribute the data, and to cite the IndoBias paper in any work that uses it.

IndoBias

Data for IndoBias: A Dual Track Culturally Grounded Benchmark for LLMs Bias Evaluation in Indonesian Languages (arXiv:2606.01260).

Code: github.com/IkhlasulHanif/IndoBias

Warning: this dataset contains stereotypical statements about social groups that may be offensive, harmful or biased. They exist to measure bias, not to endorse it.

Access

This dataset is gated. Fill in the form on this page, and the authors will review your request by hand. For questions, contact Ikhlasul Akmal Hanif (ikhlasul.hanif@mbzuai.ac.ae) or Muhammad Falensi Azmi (muhammad.azmi@mbzuai.ac.ae).

Pair-based track (`pair/`)

544 contrastive sentence pairs in each of Indonesian, Javanese, Sundanese and Makassar. The four files are row-aligned: row i is the same pair in each language.

Column	Description
`protrope`	The prototypical (stereotype-consistent) sentence
`antitrope`	The same sentence with the attribute flipped
`dimension`	Fine-grained bias dimension (original Indonesian label)
`dimension_en`	Fine-grained bias dimension (English label)
`domain`	One of five domains (original Indonesian label)
`domain_en`	One of five domains (English label)

`domain_en`	`domain`	Pairs
Economic Status	Status Ekonomi	241
Identity and Demographics	Identitas dan Demografi	93
Cultural and Geographic	Kultural dan Geografis	93
Social and Family Roles	Peran Sosial dan Keluarga	68
Ideology and Religion	Ideologi dan Agama	49

`dimension_en`	`dimension`	Pairs
occupation	pekerjaan	123
education	pendidikan	91
socio_territorial_region	kawasan_sosio_teritorial	53
ethnicity	ethnicity	46
family_role	peran_keluarga	35
gender	gender	29
religion	religion	25
generation	generasi	18
rural_urban_divide	rural_urban_divide	18
social_status	status_sosial	17
marital_status	status_pernikahan	16
social_class	kelas_sosial	15
access_and_infrastructure	akses_dan_infrastruktur	12
economy	ekonomi	12
ideology	ideologi	12
political_affiliation	afiliasi_politik	12
cultural_affiliation	afiliasi_budaya	6
domicile	domisili	4

A model is biased toward a pair if it assigns lower perplexity to protrope than to antitrope.

Generation-based track (`qa/groups/`)

The Indonesian group lists used to fill the generation prompts, with columns group_id, group_name and notes:

File	Groups
`ethnicity_indonesia.csv`	81
`media.csv`	84
`name.csv`	78
`university.csv`	68
`lembaga.csv` (government institutions)	60
`organisasi_bisnis.csv` (business organizations)	49
`religion.csv`	29
`political_party.csv`	20
`gender.csv`	2

The prompt templates and the SPI, O*NET and WGI stereotype dimensions are in the code repository (indobias_qa/tasks.py and indobias_qa/data/dimensions.py).

Usage

from datasets import load_dataset

pairs = load_dataset("ikhlasulakmalh/IndoBias", "pair_id", split="train")
groups = load_dataset("ikhlasulakmalh/IndoBias", "qa_groups", split="train")

License

CC BY-NC-SA 4.0. The code is MIT-licensed separately in the GitHub repository.

Citation

@misc{hanif2026indobias,
  title         = {IndoBias: A Dual Track Culturally Grounded Benchmark for LLMs Bias Evaluation in Indonesian Languages},
  author        = {Hanif, Ikhlasul Akmal and Azmi, Muhammad Falensi and Tjiaranata, Filbert Aurelian and Yulianrifat, Eryawan Presma and Koto, Fajri},
  year          = {2026},
  eprint        = {2606.01260},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2606.01260}
}

Downloads last month: 31

Total file size:

498 kB

Paper for ikhlasulakmalh/IndoBias

IndoBias: A Dual Track Culturally Grounded Benchmark for LLMs Bias Evaluation in Indonesian Languages

Paper • 2606.01260 • Published May 31 • 2

Request access to IndoBias