首页 | 本学科首页   官方微博 | 高级检索  
     

基于注意力机制的陆空通话语义感知方法
引用本文:乔璇,郭晓静,杨金锋.基于注意力机制的陆空通话语义感知方法[J].计算机应用与软件,2022,39(1):212-218.
作者姓名:乔璇  郭晓静  杨金锋
作者单位:中国民航大学计算机科学与技术学院 天津 300300;中国民航大学 天津 300300;深圳职业技术学院 广东 深圳 518055
摘    要:民航运输过程中陆空通话双方出现信息不一致的问题会威胁飞行安全.为了保障陆空通话的准确性,提出一种基于注意力机制的陆空通话语义感知模型,对陆空通话复诵与问答进行语义感知与判别.该模型结合多种特征向量表示句子并对句子向量进行编码,建立注意力交互矩阵计算指令与答复的软对齐结果,采用前馈神经网络和增强差异性的方法进行语义信息的...

关 键 词:陆空通话  语义感知  句子匹配  注意力机制

SEMANTIC PERCEPTION OF AVIATION RADIOTELEPHONY COMMUNICATION BASED ON ATTENTION MECHANISM
Qiao Xuan,Guo Xiaojing,Yang Jinfeng.SEMANTIC PERCEPTION OF AVIATION RADIOTELEPHONY COMMUNICATION BASED ON ATTENTION MECHANISM[J].Computer Applications and Software,2022,39(1):212-218.
Authors:Qiao Xuan  Guo Xiaojing  Yang Jinfeng
Affiliation:(College of Computer Science and Technology,Civil Aviation University of China,Tianjin 300300,China;Civil Aviation University of China,Tianjin 300300,China;Shenzhen Polytechnic,Shenzhen 518055,Guangdong,China)
Abstract:The inconsistent information between the two sides of radiotelephony communication during civil aviation transportation will threaten flight safety.To ensure the accuracy of radiotelephony communication,we propose a semantic perception model based on the attention mechanism to semantically perceive and discriminate the repetition and question answering of radiotelephony communication.The model combined multiple feature vectors to represent sentences and encode the sentence vectors.It established the attention interaction matrix to calculate the soft alignment results of instructions and responses,and used feedforward neural network and the method of enhancing differentiation to compare and aggregate semantic information,so as to achieve semantic matching and perception.The experimental results show that the test accuracy of the proposed method in radiotelephony communication dataset reaches 94.57%,which is more suitable for semantic analysis in aviation radiotelephony communication than other matching models.
Keywords:Aviation radiotelephony communication  Semantic perception  Sentence matching  Attention mechanism
本文献已被 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号