tldc/mslr2022

多文档文献综述摘要（MSLR）共享任务数据集旨在研究如何从不同的临床研究中总结医学证据用于文献综述。综述为临床护理提供了最高质量的证据，但手动制作成本高昂。通过自然语言处理（NLP）的（半）自动化可以在不牺牲严谨性的前提下加速证据合成。MSLR共享任务使用两个数据集来评估当前多文档摘要技术在此任务中的状态，并鼓励在此领域开发模型贡献、支架任务、模型可解释性方法以及改进的自动评估方法。数据集包含结构化数据实例，如综述ID、PMID、标题和临床研究的摘要，支持摘要和文本生成任务。

The Multidocument Summarization for Literature Review (MSLR) Shared Task aims to study how medical evidence from different clinical studies are summarized in literature reviews. Reviews provide the highest quality of evidence for clinical care, but are expensive to produce manually. (Semi-)automation via NLP may facilitate faster evidence synthesis without sacrificing rigor. The MSLR shared task uses two datasets to assess the current state of multidocument summarization for this task, and to encourage the development of modeling contributions, scaffolding tasks, methods for model interpretability, and improved automated evaluation methods in this domain. The dataset includes structured data instances with review IDs, PMIDs, titles, and abstracts from clinical studies, supporting tasks like summarization and text2text-generation.