VizNet
收藏资源简介:
VizNet是由麻省理工学院媒体实验室创建的一个大规模数据集,包含超过3100万个数据集,主要从开放数据存储库和在线可视化画廊中编译而成。这些数据集平均包含17条记录,跨越3个维度,其中51%的维度记录分类数据,44%记录定量数据,仅有5%记录时间数据。VizNet的创建旨在为比较可视化设计技术提供必要的共同基准,并开发用于自动化视觉分析的基准模型和算法。该数据集适用于大规模在线众包实验,特别是在评估用户任务和数据分布对可视化编码有效性的影响方面。
VizNet is a large-scale dataset created by the MIT Media Lab, containing over 31 million datasets compiled primarily from open data repositories and online visualization galleries. On average, each of these datasets contains 17 records across 3 dimensions, among which 51% of the dimensions hold categorical data, 44% hold quantitative data, and only 5% hold temporal data. VizNet was developed to provide the necessary common benchmarks for comparing visualization design techniques, as well as to develop benchmark models and algorithms for automated visual analysis. This dataset is suitable for large-scale online crowdsourcing experiments, particularly in evaluating the impact of user tasks and data distributions on the effectiveness of visualization encoding.

- 1VizNet: Towards A Large-Scale Visualization Learning and Benchmarking Repository麻省理工学院媒体实验室 · 2019年



