Gong Jie, Cao Yang, Zijing Miao, Chen Qiaosen
School of Computer Science, South China Normal University, Guangzhou, Guangdong, China.
PeerJ Comput Sci. 2023 Mar 13;9:e1275. doi: 10.7717/peerj-cs.1275. eCollection 2023.
Event Extraction (EE) is an essential and challenging task in information extraction. Most existing event extraction methods do not specifically target the Chinese geological hazards domain. This is due to the unique characteristics of the Chinese language and the lack of Chinese geological hazard datasets. To address these challenges, we propose a novel multi-word lexical feature enhancement framework (MFEE). It effectively implements Chinese event extraction in the geological hazard domain by introducing lexical information and the designed lexical feature weighting decision method. In addition, we construct a large-scale Chinese geological hazard dataset (CGHaz). Experimental results on this dataset and the ACE 2005 dataset demonstrate the approach's effectiveness. The datasets can be found at https://github.com/JieGong1130/MFEE-dataset. The code can be found at https://github.com/JieGong1130/MFEE-master.
事件抽取（EE）是信息抽取中一项重要且具有挑战性的任务。大多数现有的事件抽取方法并未专门针对中国地质灾害领域。这是由于中文语言的独特特性以及缺乏中文地质灾害数据集所致。为应对这些挑战，我们提出了一种新颖的多词词汇特征增强框架（MFEE）。通过引入词汇信息和所设计的词汇特征加权决策方法，它有效地实现了地质灾害领域的中文事件抽取。此外，我们构建了一个大规模的中文地质灾害数据集（CGHaz）。在该数据集和ACE 2005数据集上的实验结果证明了该方法的有效性。数据集可在https://github.com/JieGong1130/MFEE-dataset找到。代码可在https://github.com/JieGong1130/MFEE-master找到。
