Design And Implementation Of Quality Diagnostic System For Chinese-English Text Translation

As an international language,English is the most important second language in all learning stages in China.Translation is a necessary skill in bilingual learning,which tests the learner’s knowledge of words,grammar and syntactic structure.However,in practical teaching,the speed and reliability of teachers’ feedback will directly affect the improvement of learners’ translation ability.Therefore,with the help of natural language processing and network technology,designing and realizing automatic marking of translation subjective questions can not only reduce teachers’ marking burden,but also promote the automation of centralized marking process of large-scale examinations.This paper takes the automatic marking of Chinese-English translation in the form of text as the main research object,aiming at realizing an Internet-based automatic diagnosis system for Chinese-English translation quality for Chinese English learners.Specifically,from the perspective of translation evaluation,this paper not only evaluates the shallow expression forms of the translation,but also explores the deep semantics of the translation,and designs a comprehensive diagnostic model for the quality of Chinese-English translation.Then,considering the actual application requirements of large-scale examination paper marking,a server architecture is constructed to meet the requirements of high concurrency and high reliability.The main research contents of this paper are as follows:1.A text semantic similarity analysis method KPTM(Knowledge Prior Topic Model)is studied and designed.In this method,a priori knowledge is introduced into the topic model to analyze the impact of concepts on semantics,and the serialized effective keywords are combined to realize the hybrid measure of concepts and keywords to calculate the similarity between texts,and to analyze the deep semantic quality of translated texts.2.This paper studies and designs a diagnostic model of Chinese-English translation quality based on the experience of manual translation marking.The model takes the semantic similarity as the main evaluation index,and combines the evaluation of translation expression form,including spelling correction and word shape matching,to achieve the multidimensional quality diagnosis of translation.3.This paper studies the NIO network model and the master-slave reactor multithreading mode,and designs an asynchronous,event-driven translation receiving and correcting processing architecture to achieve high concurrency and high reliability translation automatic quality diagnosis.The architecture integrates multiple modules for custom business logic processing and optimizes traditional file transfers based on direct memory mapping.Experimental data of protocol test and load test show that the transmission performance of this architecture is about 36% higher than that of traditional file transmission,and it supports the concurrent correction of 200,000 English translations,which meets the requirements of practical applications.4.Finally,this paper combines the high-concurrent correction framework with the Chinese-English translation quality diagnosis model,and finally realizes an Internet-based Chinese-English translation quality diagnosis system for Chinese English learners to achieve accurate and efficient automatic translation evaluation.