My-Custom-AI/ACE-BB

ACE-BB（Agentic e-Commerce - 选择行为与偏见）是一个用于研究自主AI代理在电子商务环境中购物的行为的实验集。该数据集通过提供合成任务，对代理的选择行为和偏见进行评估，从而可以进行控制的实验研究AI的购物行为。数据集包含两个子集：choice_behavior和market_share，每个子集包含500个针对每个产品类别的随机场景，场景中外部变量包括产品位置、平台标签（赞助、总体选择、稀缺）和产品属性（价格、评分、评论数量），用于测量AI代理对平台操作杆和产品特征的敏感度。

The ACE-BB (Agentic e-Commerce - choice Behavior & Biases) dataset is a set of experiments designed to study the behavior of autonomous AI agents when shopping in e-commerce settings. The dataset enables controlled experiments on AI shopping behavior by providing synthetic tasks that evaluate agents choice behavior and biases. It includes two subsets: choice_behavior and market_share, each containing 500 randomized scenarios per product category, with exogenous variation in product positions, platform tags (Sponsored, Overall Pick, Scarcity), and product attributes (price, rating, number of reviews) to measure AI agent sensitivity to platform levers and product characteristics.