基于情感语义增强编解码的神经机器翻译方法

Neural Machine Translation Method Based on Emotional Semantics Enhanced Encoding and Decoding

针对目前神经机器翻译模型仅依赖平行语料训练而无法充分挖掘深层语言知识的问题,提出一种基于情感语义增强编解码的神经机器翻译方法,旨在通过引入额外的情感语义,提高模型对语言深层次信息的理解能力。首先,利用word2vec技术获取语料中所有单词的词嵌入,将其输入到一个融合模型中进行训练。该融合模型结合了基于GRU和文档嵌入的机制,以获取单词级别和文档级别的情感语义表征;其次,在情感融合阶段,采用加权公式将单词级别和文档级别的情感语义有机地融合,形成更为综合的情感语义表征;最后,将此表征与上下文语义表征按位相加,以全面引入情感信息,并将其作为输入传递到机器翻译模型的编码器和解码器中。在多个基准数据集上的实验显示,相较于传统的Transformer模型,该方法在IWSLT数据集上性能显著提升,BLEU值增加1.3至1.62。在WMT数据集上也取得良好性能,证实了融合情感语义在机器翻译中的有效性。

To address the problem that current neural machine translation models rely solely on parallel corpus training and cannot fully tap into deep linguistic knowledge,we propose a neural machine translation method based on emotional semantic enhancement coding and decoding to improve the model’s ability to understand deep linguistic information by introducing additional emotional semantics.Firstly,word2vec technology is used to obtain word embeddings for all words in the corpus,which are then input into a fusion model for training.This fusion model combines mechanisms based on GRU and document embedding to obtain emotional semantic representations at the word and document levels.Secondly,in the emotional fusion stage,a weighted formula is used to integrate the emotional semantic representations at the word and document levels organically,forming a more comprehensive emotional semantic representation.Finally,this representation is added bitwise with contextual semantic representations to fully introduce emotional information,and passed as input to the encoder and decoder of the machine translation model.Experiments on multiple benchmark datasets show that compared to traditional Transformer models,the proposed method significantly improves performance on the IWSLT dataset,with BLEU values increasing by 1.3 to 1.62.It also achieves good performance on the WMT dataset,confirming the effectiveness of integrating emotional semantics in machine translation.