FREDSum

FREDSum是一个专为法语政治辩论摘要设计的数据集，由法国林果科技创建。该数据集包含142个手动转录和注释的政治辩论，覆盖广泛的主题和观点。数据集的创建过程涉及专家的转录和注释，旨在为对话摘要模型提供高质量的训练和评估数据。FREDSum的应用领域包括提升法语及多语言对话摘要技术，使更广泛的受众能够以母语参与重要讨论。FREDSum is a dataset specifically tailored for French political debate summarization, developed by LinGuo Technology France. It contains 142 manually transcribed and annotated political debates covering a diverse range of topics and perspectives. The development of FREDSum involved expert-led transcription and annotation work, with the core objective of providing high-quality training and evaluation data for conversational summarization models. Its application areas include advancing French and multilingual conversational summarization technologies, and enabling a broader audience to participate in important discussions in their native language.