Multi-Narrative Semantic Overlap (MNSO) Benchmark Dataset
收藏arXiv2022-01-14 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2201.05294v1
下载链接
链接失效反馈官方服务：
资源简介：
MNSO基准数据集由奥本大学BDI实验室创建，包含2925对来自不同政治立场的新闻文章，旨在研究多叙事语义重叠问题。数据集通过网络爬虫从AllSides.com获取，并经过人工标注，形成了411个不同的语义重叠基准。该数据集的应用领域包括信息检索、问答系统和翻译质量评估，旨在通过自动化技术提高对多视角信息的理解和处理效率。
The MNSO benchmark dataset was developed by the BDI Laboratory at Auburn University. It contains 2,925 pairs of news articles spanning diverse political stances, and is designed to study the problem of multi-narrative semantic overlap. Collected from AllSides.com through web crawling, the dataset was subsequently manually annotated, yielding 411 distinct semantic overlap benchmarks. Its application domains include information retrieval, question answering systems, and translation quality assessment, with the core objective of enhancing the efficiency of understanding and processing multi-perspective information via automated technologies.
提供机构：
奥本大学BDI实验室
创建时间：
2022-01-14
