HunSum-1
收藏arXiv2023-02-01 更新2024-06-21 收录
下载链接：
https://github.com/dorinapetra/summarization
下载链接
链接失效反馈官方服务：
资源简介：
HunSum-1是一个专为匈牙利语抽象摘要任务设计的数据集，由布达佩斯技术与经济大学和计算机科学与控制研究所联合创建。该数据集包含114万篇新闻文章，这些文章是从9个主要的匈牙利新闻网站通过CommonCrawl收集、清洗和去重处理得到的。数据集的创建过程涉及复杂的网页抓取和数据预处理技术。HunSum-1数据集主要用于训练和评估抽象摘要模型，特别是在处理匈牙利语内容时，旨在提高摘要的准确性和相关性。
HunSum-1 is a dataset specifically designed for Hungarian abstractive summarization tasks, jointly created by the Budapest University of Technology and Economics and the Institute for Computer Science and Control. The dataset contains 1.14 million news articles collected, cleaned and deduplicated from nine major Hungarian news websites via CommonCrawl. The construction of HunSum-1 involves sophisticated web scraping and data preprocessing techniques. It is primarily used for training and evaluating abstractive summarization models, particularly for Hungarian-language content, with the goal of improving the accuracy and relevance of generated summaries.
提供机构：
布达佩斯技术与经济大学自动化与应用信息学系 2计算机科学与控制研究所 厄特沃什·罗兰研究网络
创建时间：
2023-02-01
