Hakka Cultural Knowledge Cognitive Domain Dataset
收藏资源简介:
Hakka Cultural Knowledge Cognitive Domain Dataset是由国立联合大学开发的一个综合性数据集,旨在评估大型语言模型(LLMs)在理解和处理客家文化知识方面的表现。该数据集包含10,158个问题,涵盖了客家文化的各个方面,如语言、建筑、饮食和节日等。数据集的创建过程结合了Bloom’s Taxonomy的六个认知层次和Retrieval-Augmented Generation(RAG)技术,确保了对LLMs在文化认知任务中的全面评估。该数据集主要应用于AI驱动的文化知识保存和传播领域,旨在解决文化误解和偏见问题,促进多元文化的准确传播。
Hakka Cultural Knowledge Cognitive Domain Dataset is a comprehensive dataset developed by National United University, designed to assess the performance of Large Language Models (LLMs) in comprehending and processing Hakka cultural knowledge. The dataset contains 10,158 questions covering all dimensions of Hakka culture, including language, architecture, cuisine, festivals and other related aspects. The construction of this dataset integrates the six cognitive levels of Bloom’s Taxonomy and Retrieval-Augmented Generation (RAG) technology, ensuring a comprehensive evaluation of LLMs in cultural cognitive tasks. This dataset is primarily applied in the field of AI-driven cultural knowledge preservation and dissemination, with the goals of resolving cultural misunderstandings and biases, and promoting the accurate dissemination of diverse cultures.




