NewsInterview
收藏资源简介:
NewsInterview数据集由加州大学伯克利分校和南加州大学的研究团队创建,包含40,000条来自NPR和CNN的两人信息访谈记录。该数据集旨在研究大型语言模型在新闻采访中的基础语言和战略对话能力。数据集通过收集和处理公开的采访记录创建,涵盖了丰富的自然语言交互场景。其应用领域主要集中在提升大型语言模型在多轮对话中的策略性和说服性对话能力,以解决模型在实际应用中的不足。
NewsInterview Dataset was developed by a research team from the University of California, Berkeley and the University of Southern California, encompassing 40,000 two-person news interview transcripts sourced from NPR and CNN. The primary goal of this dataset is to investigate the fundamental linguistic and strategic conversational capabilities of large language models (LLMs) within the context of news interviews. It is constructed through the collection and processing of publicly available interview records, covering a wide range of natural language interaction scenarios. Its core application areas focus on enhancing the strategic and persuasive conversational abilities of large language models in multi-turn dialogues, so as to address the shortcomings of such models in real-world applications.




