---
dataset_info:
- config_name: nb
  features:
  - name: id
    dtype: string
  - name: question
    dtype: string
  - name: choices
    struct:
    - name: label
      sequence: string
    - name: text
      sequence: string
  - name: answer
    dtype: string
  - name: quiz
    dtype: string
  - name: curated
    dtype: bool
  splits:
  - name: test
    num_bytes: 922278
    num_examples: 3600
  download_size: 492918
  dataset_size: 922278
- config_name: nn
  features:
  - name: id
    dtype: string
  - name: question
    dtype: string
  - name: choices
    struct:
    - name: label
      sequence: string
    - name: text
      sequence: string
  - name: answer
    dtype: string
  - name: quiz
    dtype: string
  - name: curated
    dtype: bool
  splits:
  - name: test
    num_bytes: 340107
    num_examples: 1330
  download_size: 175131
  dataset_size: 340107
configs:
- config_name: nb
  data_files:
  - split: test
    path: nb/test-*
- config_name: nn
  data_files:
  - split: test
    path: nn/test-*
license: mit
task_categories:
- question-answering
language:
- nb
- nn
pretty_name: NRK-Quiz-QA
size_categories:
- 1K<n<10K
---

# Dataset Card for NRK-Quiz-QA

## Dataset Details

### Dataset Description

NRK-Quiz-QA is a multiple-choice question answering (QA) dataset designed for zero-shot evaluation of language models' Norwegian-specific and world knowledge. It comprises 4.9k examples from over 500 quizzes on Norwegian language and culture, spanning both written standards of Norwegian: Bokmål and Nynorsk (the minority variant). These quizzes are sourced from NRK, the national public broadcaster in Norway. Each example includes a question and 2 to 5 answer choices.

NRK-Quiz-QA is part of the collection of Norwegian QA datasets, which also includes: [NorOpenBookQA](https://huggingface.co/datasets/ltg/noropenbookqa), [NorCommonSenseQA](https://huggingface.co/datasets/ltg/norcommonsenseqa), [NorTruthfulQA (Multiple Choice)](https://huggingface.co/datasets/ltg/nortruthfulqa_mc), and [NorTruthfulQA (Generation)](https://huggingface.co/datasets/ltg/nortruthfulqa_gen). We describe our high-level dataset creation approach here and provide more details, general statistics, and model evaluation results in our paper.

- **Curated by:** The [Language Technology Group](https://www.mn.uio.no/ifi/english/research/groups/ltg/) (LTG) at the University of Oslo
- **Shared by:** NRK through initial contact by Språkbanken.
- **Language:** Norwegian (Bokmål and Nynorsk)
- **Repository:** [github.com/ltgoslo/norqa](https://github.com/ltgoslo/norqa)
- **Paper:** [arxiv.org/abs/2501.11128](https://arxiv.org/abs/2501.11128) (to be presented at NoDaLiDa/Baltic-HLT 2025)
- **License:** MIT


### Citation

```
@article{mikhailov2025collection,
  title={A Collection of Question Answering Datasets for Norwegian},
  author={Mikhailov, Vladislav and M{\ae}hlum, Petter and Lang{\o}, Victoria Ovedie Chruickshank and Velldal, Erik and {\O}vrelid, Lilja},
  journal={arXiv preprint arXiv:2501.11128},
  year={2025}
}
```

### Uses

NRK-Quiz-QA is intended to be used for zero-shot evaluation of language models for Norwegian. 


## Dataset Creation

NRK-Quiz-QA is based on a collection of quizzes from between the years of 2017 and 2024, provided by NRK. The quiz data is of high quality, but we perform a targeted adaptation to ensure correct time references. This annotation stage is performed by three students in linguistics and computer science, all native Norwegian speakers. Two annotators work on Bokmål, and one annotator works on Nynorsk. There are three main curation aspects:

1. **Temporal adjustment**: The annotators adjust temporal references to fit the current time.
2. **Content filtering**: The annotators discard examples requiring images or sounds for answering.
3. **Data cleaning**: The annotators remove unnecessary text segments (e.g., web page artifacts), and irrelevant content in the questions (e.g., comments that guide the user through the quiz).


#### Personal and Sensitive Information

The data does not contain information considered personal or sensitive.


## Dataset Structure

### Dataset Instances

Each dataset instance looks as follows:

#### Bokmål

```
{
    'id': '1.16235578-6',
    'question': 'Juletrefester er moro for kidsa. I sin tid bidro festene også til å spre den juletreskikken vi kjenner i dag. Vet du i hvilket århundre juletreet meldte sin ankomst i Norge?',
    'choices': {
        'label': ['A', 'B', 'C'],
        'text': ['1900-tallet', '1800-tallet', '1700-tallet']
    },
    'answer': 'B',
    'quiz': 'Romjulsquiz',
    'curated': True
}
```

#### Nynorsk

```
{
    'id': '1.14319825-6b',
    'question': 'Sogndal spelte i førstedivisjon i 2019, etter at dei tapte 3–1 borte mot Aalesund i kvalifiseringskampen. Men kven scora Sogndal sitt mål?',
    'choices': {
        'label': ['A', 'B', 'C', 'D'],
        'text': [
            'Ulrik Flo',
            'Sigurd Haugen',
            'Ulrik Fredriksen',
            'Eirik Birkelund'
        ]
    },
    'answer': 'C',
    'quiz': 'Helgequiz frå NRK Sogn og Fjordane',
    'curated': True
}
```

### Dataset Fields

`id`: an example id \
`question`: a quiz question \
`choices`: answer choices (`label`: a list of labels; `text`: a list of possible answers) \
`answer`: the correct answer from the list of labels (A/B/C/D/E) \
`quiz`: a quiz name \
`curated`: an indicator of whether an example has been curated or not


## Dataset Card Contact

* Vladislav Mikhailov (vladism@ifi.uio.no)
* Lilja Øvrelid (liljao@ifi.uio.no)