Synthetic Constructive Benchmark

该数据集是一组为评估建设性领域中的偏好征询方法而生成的合成用户偏好数据。数据集中包含两类用户，他们的偏好向量分别从均匀分布和正态分布中抽取。可行配置的空间是r个属性笛卡尔积，每个属性取值范围为[r]。该数据集用于评估选择感知器算法与其他最先进方法的有效性。其规模包括256个产品和16个特征（当r=4时）。任务是对偏好进行征询。

This dataset is a collection of synthetic user preference data generated to evaluate preference elicitation methods in the constructive domain. The dataset contains two types of users, whose preference vectors are sampled from a uniform distribution and a normal distribution respectively. The space of feasible configurations is the Cartesian product of r attributes, where each attribute takes values within the set [r]. This dataset is utilized to assess the effectiveness of the Choice Perceptron algorithm and other state-of-the-art methods. When r=4, the dataset consists of 256 products and 16 features. The core task of this dataset is preference elicitation.