README.md

metadata

license: openrail
language:
  - en
tags:
  - web agent
  - multimodal
dataset_info:
  features:
    - name: action_uid
      dtype: string
    - name: raw_html
      dtype: string
    - name: cleaned_html
      dtype: string
    - name: operation
      dtype: string
    - name: pos_candidates
      sequence: string
    - name: neg_candidates
      sequence: string
    - name: website
      dtype: string
    - name: domain
      dtype: string
    - name: subdomain
      dtype: string
    - name: annotation_id
      dtype: string
    - name: confirmed_task
      dtype: string
    - name: screenshot
      dtype: image
    - name: action_reprs
      sequence: string
  splits:
    - name: test_website
      num_bytes: 1589513606.713
      num_examples: 1019
    - name: test_task
      num_bytes: 2004628575.972
      num_examples: 1339
    - name: test_domain
      num_bytes: 5128899015.440001
      num_examples: 4060
  download_size: 3172992799
  dataset_size: 8723041198.125

Dataset Description

• Homepage: https://osu-nlp-group.github.io/SeeAct/
• Repository: https://github.com/OSU-NLP-Group/SeeAct
• Paper: https://arxiv.org/abs/2401.01614
• Point of Contact: Boyuan Zheng

Dataset Summary

Multimodal-Mind2Web is the multimodal version of Mind2Web, a dataset for developing and evaluating generalist agents for the web that can follow language instructions to complete complex tasks on any website. In this dataset, we align each HTML document in the dataset with its corresponding webpage screenshot image from the Mind2Web raw dump, which undergoes human verification to confirm element visibility and correct rendering for action prediction.

Dataset Structure

Data Splits

• train: 7775 actions from 1009 tasks.
• test:
  • Cross Task: 1339 actions from 177 tasks. Tasks from the same website are seen during training.
  • Cross Website: 1019 actions from 142 tasks. Websites are not seen during training.
  • Cross Domain: 4060 actions from 694 tasks. Entire domains are not seen during training.

The three test set splits have went through human verification to confirm element visibility and correct rendering for action prediction.

Each line in the dataset is an action consisting of screenshot image, HTML text and other fields required for action prediction, for the convenience of inference.

Data Fields

• "annotation_id" (str): unique id for each task
• "website" (str): website name
• "domain" (str): website domain
• "subdomain" (str): website subdomain
• "confirmed_task" (str): task description
• "action_reprs" (list[str]): human readable string representation of the action sequence
• "screenshot" (str): path to the webpage screenshot image corresponding to the HTML.
• "action_uid" (str): unique id for each action (step)
• "raw_html" (str): raw html of the page before the action is performed
• "cleaned_html" (str): cleaned html of the page before the action is performed
• "operation" (dict): operation to perform
  • "op" (str): operation type, one of CLICK, TYPE, SELECT
  • "original_op" (str): original operation type, contain additional HOVER and ENTER that are mapped to CLICK, not used
  • "value" (str): optional value for the operation, e.g., text to type, option to select
• "pos_candidates" (list[dict]): ground truth elements. Here we only include positive elements that exist in "cleaned_html" after our preprocessing, so "pos_candidates" might be empty. The original labeled element can always be found in the "raw_html".
  • "tag" (str): tag of the element
  • "is_original_target" (bool): whether the element is the original target labeled by the annotator
  • "is_top_level_target" (bool): whether the element is a top level target find by our algorithm. please see the paper for more details.
  • "backend_node_id" (str): unique id for the element
  • "attributes" (str): serialized attributes of the element, use json.loads to convert back to dict
• "neg_candidates" (list[dict]): other candidate elements in the page after preprocessing, has similar structure as "pos_candidates"

Disclaimer

This dataset was collected and released solely for research purposes, with the goal of making the web more accessible via language technologies. The authors are strongly against any potential harmful use of the data or technology to any party.

Citation Information

@article{zheng2024seeact,
  title={GPT-4V(ision) is a Generalist Web Agent, if Grounded},
  author={Boyuan Zheng and Boyu Gou and Jihyung Kil and Huan Sun and Yu Su},
  journal={arXiv preprint arXiv:2401.01614},
  year={2024},
}

@inproceedings{deng2023mindweb,
  title={Mind2Web: Towards a Generalist Agent for the Web},
  author={Xiang Deng and Yu Gu and Boyuan Zheng and Shijie Chen and Samuel Stevens and Boshi Wang and Huan Sun and Yu Su},
  booktitle={Thirty-seventh Conference on Neural Information Processing Systems},
  year={2023},
  url={https://openreview.net/forum?id=kiYqbO3wqw}
}