Design And Implementation Of Multilingual Chat System Based On Machine Translation

With the rapid development of computer technology and the Internet,great changes have taken place in people's daily life.The various convenient online service systems thus produced have profoundly changed people's communication and contact methods.Chat system emerged as a system to meet people's real-time communication needs,which greatly shortens the distance between users in different regions and facilitates the dissemination and acquisition of information.On the other hand,with the increasing development of internationalization,the need for international communication has emerged in different fields.However,language barriers inevitably exist in international communication.This topic aims to provide users with a multilingual chat system through language detection and machine translation technology.The user only needs to set up a common language,and the system will translate the message sent by the user into the common language of the target user and send it to the other party after the translation is completed.The user can meet the needs of multilingual communication without paying attention to the common language of the other party.For language detection technology,this paper proposes to use BPE method to solve the problem of large vocabulary of detection model,use TextCNN model to improve detection accuracy,and design experiments to compare,using the above method to improve detection accuracy.For the lack of parallel corpus in translation between minority languages and Chinese,which seriously affects the effect of translation model,this paper proposes an innovation in parallel corpus collection method,which greatly enriches the amount of parallel corpus that minority languages can collect and ensures the translation quality of minority language model.In order to ensure the translation quality of machine translation and solve the dependence of machine translation models on the quality of parallel corpora,this paper proposes a machine translation model integrating prior knowledge.The model integrates the pre-training model with the machine translation model by using the knowledge encoder structure,introduces prior knowledge,reduces the dependence of the model on parallel corpora,and the model can be used as a general model structure integrating the pre-training model with Seq2seq model,and designs experiments to compare the effects of various models.Compared with the general Transformer model,the machine translation model integrating prior knowledge improves the BLEU value.This topic first discusses the background and significance of the topic,and determines the main research content of this topic.Then the related technologies and theories involved in the research and implementation of the subject are introduced,including front-end development framework,server-side development framework,algorithm model,etc.Then,the key technologies and innovations of this topic are described in detail,namely,the research of language detection model,the innovation of parallel corpus collection method aiming at the scarcity of minority language data,and the research of machine translation model integrating prior knowledge.According to the requirements of software engineering,the requirements of the system are analyzed,including functional requirements and non-functional requirements.Based on the requirements analysis,the overall design of the system is carried out,the overall architecture and functional modules of the system are determined,and the database design,front-end page design and communication interface design are completed.Based on the overall design,the system is designed and implemented in detail according to the division of functional modules,and the timing diagram or flow chart of each module is established.Finally,functional and non-functional tests are carried out on the system,and all test results are analyzed.According to the actual needs,based on natural language processing technologies such as machine translation and Web front-end and back-end development technologies,this paper designs and implements a multilingual chat system that is simple and easy to use and meets the actual needs.