TREC Complex Answer Retrieval (TREC CAR)
收藏arXiv2017-05-13 更新2024-06-21 收录
下载链接：
http://trec-car.cs.unh.edu/
下载链接
链接失效反馈官方服务：
资源简介：
TREC CAR数据集是由TREC组织创建的，旨在解决复杂查询的段落检索问题。该数据集包含从英文维基百科中提取的700万段落，每个段落都与维基百科页面中的特定部分相关联。数据集的创建过程涉及对维基百科文章的处理，保留了具有层次结构标题的文章，并去除了信息框、图像和包装器。TREC CAR数据集的应用领域主要集中在信息检索研究，特别是针对复杂查询的答案检索，旨在通过各种技术如查询扩展和深度神经网络来提高检索性能。
The TREC CAR dataset was developed by the TREC organization to tackle the passage retrieval problem for complex queries. This dataset consists of 7 million passages extracted from English Wikipedia, where each passage is linked to a specific section of a corresponding Wikipedia page. The dataset creation process entails processing Wikipedia articles, retaining those with hierarchical section titles while removing infoboxes, images, and navigational wrappers. The TREC CAR dataset is primarily utilized in information retrieval research, especially answer retrieval for complex queries, aiming to enhance retrieval performance through various technologies such as query expansion and deep neural networks.
提供机构：
曼海姆大学
创建时间：
2017-05-13
