基于多层注意力机制与双层BiLSTM结合的民航事件关系抽取模型  

Civil Aviation Relationship Extraction Model Based on the Combination of Multi-layer Attention Mechanism and Double-layer BiLSTM

作　　者：马婷[1] 张潇峰 MA Ting;ZHANGXiaofeng(School of Computer Science,Civil Aviation Flight University of China,Guanghan Sichuan 618307,China)

机构地区：[1]中国民用航空飞行学院计算机学院,四川广汉618307

出　　处：《佳木斯大学学报（自然科学版）》2024年第11期114-117,共4页Journal of Jiamusi University：Natural Science Edition

基　　金：中国民用航空飞行学院面上项目(J2021-059)。

摘　　要：弥补民航事件关系抽取研究不足,提出对注意力机制和BiLSTM结合的关系抽取模型应用到民航领域中。首先选取民航应急事件的语料文本,对选取的语料数据进行清理,将语料库中包含的词语进行矢量化生成向量,合并并连接词语属性、实体位置等,在双层BiLSTM模型中引入,以获取关系的更高层次描述;其次,词和句子级注意力机制捕获各个单词的关键性,并借助语义信息降低噪声的关联;最后,进行softmax分类得到提取结果。实验数据表明,此方法应用在民航应急数据集中,F值达到79.3%。To make up for the lack of research on civil aviation event relationship extraction,a relationship extraction model based on the combination of attention mechanism and BiLSTM is proposed and applied to the civil aviation field.0x0E0xFF?倓0x030x0FFirst of all,select the corpus text of civil aviation emergency events,clean up the selected corpus data,vectorize the words contained in the corpus to generate vectors,merge and connect word attributes,entity locations,etc.,which are introduced into the two-layer BiLSTM model to obtain a higher-level description of the relationship.Secondly,the word and sentence-level attention mechanism captures the criticality of each word and reduces the correlation of noise with the help of semantic information.0x0E0xFF?倓0x030x0FFinally,the extraction results were obtained by softmax classification.The experimental data show that this method is applied to the civil aviation emergency data set,and the F value is up to 79.3%.

关 键 词：民航突发事件 注意力机制 BiLSTM 关系抽取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]