HOWSUMM

HOWSUMM数据集是一个大规模的查询焦点多文档摘要（qMDS）数据集，由IBM研究 - AI团队创建。该数据集旨在从多个来源生成行动指南，适用于教育和工业场景。数据集包含11,121个长摘要和84,348个短摘要，这些摘要来源于wikiHow网站的文章及其引用的来源。数据集的创建过程涉及自动方法，利用了现有的人工制作qMDS数据集的统计数据。HOWSUMM数据集的应用领域包括技术支持等，旨在从网页、知识库等来源提取相关问题解决方案，并创建易于遵循的行动指南。

HOWSUMM is a large-scale query-focused multi-document summarization (qMDS) dataset developed by the IBM Research - AI Team. This dataset is designed to generate action guides from multiple sources, applicable to educational and industrial scenarios. It contains 11,121 long summaries and 84,348 short summaries, which are derived from wikiHow articles and their cited sources. The construction of the dataset adopts automated methods that leverage statistical data from existing manually curated qMDS datasets. Its application areas include technical support and other fields, with the core goal of extracting relevant problem solutions from resources such as webpages and knowledge bases and creating easy-to-follow action guides.