DebateSum
收藏资源简介:
DebateSum是由美国国家演讲与辩论协会在7年内收集的数据构建的大型论据挖掘和摘要数据集。该数据集包含187,386条独特的证据,每条证据都有相应的论据和提取摘要。数据集的创建过程涉及数千名辩论者和教练的手动生产和标注。DebateSum主要用于支持自动辩论系统中的信息检索、摘要和分类系统,旨在解决辩论研究中缺乏特定领域训练数据的问题。
DebateSum is a large-scale argument mining and summarization dataset constructed from data collected by the National Speech & Debate Association over a seven-year period. This dataset contains 187,386 unique pieces of evidence, with each piece paired with corresponding arguments and extractive summaries. The development of this dataset involved manual production and annotation completed by thousands of debaters and coaches. DebateSum is primarily designed to support information retrieval, summarization and classification systems in automated debate systems, aiming to address the scarcity of domain-specific training data in debate research.

- 1DebateSum: A large-scale argument mining and summarization dataset俄勒冈大学 · 2020年



