ACE-BB Dataset

ACE-BB (Agentic e-Commerce - choice Behavior & Biases) is a set of experiments for studying how autonomous AI agents behave when shopping in e-commerce settings. The dataset enables controlled experiments on AI shopping behavior by providing synthetic tasks which evaluate agents' choice behavior & biases.

Paper

ArXiv: https://arxiv.org/abs/XXXX.XXXXX

Dataset Description

This dataset contains experimental data from studying how AI shopping agents behave in e-commerce settings:

Choice Behavior & Biases Experiments

• 500 randomized scenarios per product category
• Exogenous variation in:
  • Product positions (2×4 grid layout)
  • Platform tags (Sponsored, Overall Pick, Scarcity)
  • Product attributes (price, rating, number of reviews)
• Measures AI agent sensitivity to platform levers and product characteristics

Dataset Subsets

• choice_behavior
• market_share

Usage

To reproduce experiments or run on new models, please refer to the ACES codebase.

Citation

If you use this dataset in your research, please cite:

@misc{allouah2025acebb,
    author       = { Amine Allouah, Omar Besbes, Josué D Figueroa, Yash Kanoria and Akshit Kumar },
    title        = { ACE-BB (Revision 499ec41) },
    year         = 2025,
    url          = { https://huggingface.co/datasets/My-Custom-AI/ACE-BB },
    doi          = { 10.57967/hf/6124 },
    publisher    = { Hugging Face }
}

License

This dataset is licensed under the MIT License and includes publically accessible data from Amazon.com.

Contact

For questions or issues, please open a GitHub issue in the ACES repository or contact the authors through the paper correspondence information.