Syntax structure and Tree-LSTM fused Chinese-Vietnamese parallel sentence pair extraction method

• Main Authors: ZHANG YINGCHEN, ZHU HAODONG, YU ZHENGTAO, GAO SHENGXIANG
• Format: Patent
• Language: Chinese; English
• Published: 15.01.2021
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention relates to a syntactic structure and Tree-LSTM fused Chinese-Vietnamese parallel sentence pair extraction method. According to the method, firstly, Chinese-Vietnamese bilingual word vectors are pre-trained, Chinese-Vietnamese bilingual words are mapped into the same semantic space, it is considered that Chinese-Vietnamese sentence structures have differences, sentence sequence structures are converted into dependency tree structures through a dependency syntax tree, syntax structure information of sentences is captured through Tree-LSTM, the part-of-speech information of the Chinese-Vietnamese bilingual sentence is spliced into a sentence semantic vector as a feature vector, and finally the vector is input into a full connection layer to train a Chinese-Vietnamese parallel sentence pair classifier. According to the method, sentence representation rules are automatically learned in a large amount of data by utilizing a deep learning method, and the problem that a large amount of human resources n