DIALOGSUM
收藏资源简介:
DIALOGSUM是一个大规模的对话摘要数据集,由浙江大学、西湖大学、爱丁堡大学和吉林大学联合创建。该数据集包含13,460个日常生活中的多轮对话,涵盖教育、工作、医疗等多个领域。数据来源于多个公开对话语料库及英语学习网站,经过清洗和预处理,由标注者从观察者角度进行摘要。DIALOGSUM旨在解决对话摘要中的挑战,如口语表达、特殊话语结构、指代和省略等,适用于客户服务管理、医疗跟踪等实际场景,同时也为个人助手记录管理提供支持。
DIALOGSUM is a large-scale dialogue summarization dataset jointly created by Zhejiang University, Westlake University, University of Edinburgh and Jilin University. This dataset contains 13,460 multi-turn daily dialogues covering various domains such as education, work and healthcare. The data is sourced from multiple public dialogue corpora and English learning websites, and has been cleaned and preprocessed, with summaries generated by annotators from an observer's perspective. DIALOGSUM aims to tackle core challenges in dialogue summarization, including spoken expressions, unique discourse structures, anaphora and ellipsis. It is applicable to practical scenarios like customer service management and medical tracking, and also provides support for record management of personal assistants.

- 1DialogSum: A Real-Life Scenario Dialogue Summarization Dataset浙江大学 · 2021年



