README.md

---
license: cc-by-nc-4.0
dataset_info:
  features:
  - name: toy_id
    dtype: string
  - name: graph
    dtype: image
  - name: nodes
    struct:
    - name: '0'
      struct:
      - name: label
        dtype: string
    - name: '100'
      struct:
      - name: label
        dtype: string
    - name: '1'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '2'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '3'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '4'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '5'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '6'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '8'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '10'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '7'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '11'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '9'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '13'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '12'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '15'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '16'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '14'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
    - name: '17'
      struct:
      - name: checked
        dtype: bool
      - name: label
        dtype: string
  - name: edges
    list:
    - name: source
      dtype: string
    - name: target
      dtype: string
  splits:
  - name: test
    num_bytes: 4614163
    num_examples: 82
  download_size: 4602704
  dataset_size: 4614163
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test-*
language:
- en
pretty_name: assembly101-graph
size_categories:
- n<1K
---

# Dataset Card for Dataset Name

<!-- Provide a quick summary of the dataset. -->
This is the instruction task graph of the ProMQA-Assembly dataset. 

## Dataset Details

### Dataset Description

<!-- Provide a longer summary of what this dataset is. -->
This is the instruction task graph of the ProMQA-Assembly dataset. QAs are avaialble in [`promqa-assembly`](https://huggingface.co/datasets/kimihiroh/promqa-assembly).


- **Curated by:** Kimihiro Hasegawa
- **Language(s) (NLP):** English
- **License:** CC BY-NC 4.0

### Dataset Sources

<!-- Provide the basic links for the dataset. -->

- **Repository:** https://github.com/assembly-101

## Uses

<!-- Address questions around how the dataset is intended to be used. -->

### Direct Use

<!-- This section describes suitable use cases for the dataset. -->

This dataset is supposed to be used as an input for the QA benchmark dataset or for task graph construction evaluations. 

### Out-of-Scope Use

<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->

This dataset should not be used for purposes other than evaluation. 

## Dataset Structure

<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->

Each example in the dataset contains the following fields: `toy_id`, `graph`, `nodes`, `edges`.

## Dataset Creation

### Curation Rationale

<!-- Motivation for the creation of this dataset. -->

We create this dataset to evaluate multimodal models on procedural activity understanding in the form of QA. 

### Source Data

<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->

#### Data Collection and Processing

<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->

This dataset is based on the videos and annotations from [Assembly 101 (Sener et al., CVPR 2022)](https://assembly-101.github.io/).

#### Who are the source data producers?

<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->

Please refer to the [Assembly 101 paper](https://openaccess.thecvf.com/content/CVPR2022/papers/Sener_Assembly101_A_Large-Scale_Multi-View_Video_Dataset_for_Understanding_Procedural_Activities_CVPR_2022_paper.pdf).

### Annotations

<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->

Instruction task graphs are annotated.

#### Annotation process

<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->

Nodes are compiled based on coarse action labels. 
We use our own annotation interface for annotation. 
82 instruction task graphs are included, among which 78 are used in our paper. 
Each annotator read the guideline before starting annotation. 
the inter-annotator agreement for edges is 0.74, and that for fastening is 0.89.

#### Who are the annotators?

<!-- This section describes the people or systems who created the annotations. -->

People who have CS graduate degrees or are in their CS graduate programs annotated data. Some of them are native English speakers, and the others are ESL speakers.

#### Personal and Sensitive Information

<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->

We did not find any private information. 

## Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

The source videos use take-apart toys available in the western countries, which may not be the same in other countries. 

### Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Users should be made aware of the risks, biases, and limitations of the dataset.

## Citation

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

```bib
@misc{hasegawa-etal-2025-promqa-assembly,
      title={ProMQA-Assembly: Multimodal Procedural QA Dataset on Assembly},
      author={Hasegawa, Kimihiro and Imrattanatrai, Wiradee and Asada, Masaki and Holm, Susan and Wang, Yuran and Zhou, Vincent and Fukuda, Ken and Mitamura, Teruko},
      year={2025},
      url={https://github.com/kimihiroh/promqa-assembly},
}
```

<!-- **APA:**

TBU -->

<!-- ## Glossary [optional]

<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->

<!-- [More Information Needed] -->

<!-- ## More Information [optional]

[More Information Needed] -->

## Dataset Card Authors

Kimihiro Hasegawa

## Dataset Card Contact

[email protected]