Abstract
"叙实性推理是机器理解文本隐含事实的关键能力之一,核心在于结合动词的语义判断动词宾语命题的真值。本研究基于首届中文叙实性推理评测任务4(FIE2025)开展叙实性推理研究,经过前期对不同模型的测验和比对,选择了Deepseek-R1模型为基座模型。提示语的总体撰写思路是:首先将动词叙实性进行分类,从传统的三分法扩展至五分法(叙实、弱叙实、反叙实、非叙实、半叙实),同时,对自然语料与人造语料进行差异化处理,再针对部分语义复杂的动词编写更加细致的判断规则。最终结果显示,自然语料的正确率达到0.9155,人造语料的正确率为0.9541,总正确率达到0.9261。"
- Anthology ID:
- 2025.ccl-2.11
- Volume:
- Proceedings of the 24th China National Conference on Computational Linguistics (CCL 2025)
- Month:
- August
- Year:
- 2025
- Address:
- Jinan, China
- Editors:
- Hongfei Lin, Bin Li, Hongye Tan
- Venue:
- CCL
- SIG:
- Publisher:
- Chinese Information Processing Society of China
- Note:
- Pages:
- 96–104
- Language:
- URL:
- https://aclanthology.org/2025.ccl-2.11/
- DOI:
- Bibkey:
- Cite (ACL):
- Zhangxiaoyi Zhangxiaoyi, 鲁嘉琪 鲁嘉琪, Zhang Da, Xiaoyu Chen, and 卢达威 卢达威. 2025. CCL25-Eval任务4系统报告:基于叙实性分类和语境特征的大语言模型叙实性推理. In Proceedings of the 24th China National Conference on Computational Linguistics (CCL 2025), pages 96–104, Jinan, China. Chinese Information Processing Society of China.
- Cite (Informal):
- CCL25-Eval任务4系统报告:基于叙实性分类和语境特征的大语言模型叙实性推理 (Zhangxiaoyi et al., CCL 2025)
- Copy Citation:
- PDF:
- https://aclanthology.org/2025.ccl-2.11.pdf