Chinese Time Reasoning (CTM)
收藏资源简介:
本研究构建了一个名为Chinese Time Reasoning (CTM)的基准数据集,该数据集由东南大学和阿里巴巴集团共同开发,旨在评估大型语言模型在时间推理方面的能力。CTM数据集包含了超过4,700个实体,涵盖了人物、地点、典故、食材和无形文化遗产等类别,时间跨度从先秦到清朝。数据集通过结构化的知识库和丰富的历史背景,为时间推理任务提供了丰富的上下文信息。
This study constructs a benchmark dataset named Chinese Time Reasoning (CTM), co-developed by Southeast University and Alibaba Group, aiming to evaluate the temporal reasoning capabilities of large language models. The CTM dataset includes over 4,700 entities covering categories such as people, locations, allusions, food ingredients, and intangible cultural heritage, with a temporal span ranging from the Pre-Qin Period to the Qing Dynasty. It provides abundant contextual information for temporal reasoning tasks via structured knowledge bases and rich historical backgrounds.

- 1Benchmarking Temporal Reasoning and Alignment Across Chinese Dynasties东南大学 · 2025年



