NLAS-MULTI
收藏资源简介:
NLAS-MULTI是由邓迪大学论证技术中心创建的多语言论证方案语料库,包含3810个自动生成的自然语言论证方案。该数据集涵盖20种不同的论证结构,涉及50个主题和两种立场(支持与反对),旨在解决论证挖掘、生成和分析领域中数据复杂性、规模限制和语言多样性问题。数据集的创建过程涉及使用GPT-3.5-TURBO和GPT-4等大型生成语言模型,通过两轮迭代生成和专家验证确保数据质量。NLAS-MULTI的应用领域广泛,可用于自动识别论证方案、深化论证分析的细节层次,以及作为未来论证挖掘研究的基准。
NLAS-MULTI is a multilingual argumentation scheme corpus developed by the Centre for Argument Technology at the University of Dundee, containing 3,810 automatically generated natural language argumentation schemes. It covers 20 distinct argumentation structures, spans 50 topics, and includes two stances (pro and con), aiming to address the challenges of data complexity, scale constraints, and linguistic diversity in the domains of argumentation mining, generation, and analysis. The construction of this dataset utilized large generative language models including GPT-3.5-TURBO and GPT-4, and ensured data quality via two rounds of iterative generation and expert validation. NLAS-MULTI has broad application prospects, which can be used for the automatic identification of argumentation schemes, the enhancement of the granularity of argumentation analysis, and as a benchmark for future argumentation mining research.




