OAEI-LLM
收藏资源简介:
OAEI-LLM数据集是由澳大利亚国立大学和莫纳什大学联合创建的,旨在评估大型语言模型(LLMs)在语义网领域中的本体匹配任务中的幻觉现象。该数据集是Ontology Alignment Evaluation Initiative (OAEI)数据集的扩展版本,包含了对LLMs在本体匹配任务中产生的幻觉进行分类和记录的新模式。数据集的创建过程包括使用LLM生成的对齐文件与原始人类标注结果进行比较,识别并分类不同的幻觉类型。OAEI-LLM数据集的应用领域主要集中在理解和改进LLMs在本体匹配任务中的表现,旨在解决LLMs在特定领域任务中产生的幻觉问题。
The OAEI-LLM dataset was jointly created by The Australian National University and Monash University, aiming to evaluate the hallucination phenomena of Large Language Models (LLMs) in ontology matching tasks within the semantic web domain. This dataset is an extended version of the Ontology Alignment Evaluation Initiative (OAEI) dataset, and it includes a new paradigm for classifying and recording the hallucinations generated by LLMs during ontology matching tasks. The dataset creation process involves comparing alignment files generated by LLMs with original human-annotated results to identify and categorize different types of hallucinations. The application scope of the OAEI-LLM dataset primarily focuses on understanding and improving the performance of LLMs in ontology matching tasks, with the goal of addressing the hallucination issues of LLMs in domain-specific tasks.




