The Design And Implementation Of Test System For Expressive Skills Based On Microservices And Containers

Expressive ability is a prominent feature of comprehensive quality.Good expres-sive language skills can help us get a head start in work,and they are good for interper-sonal communication and relationship between family members.Therefore,there is a need for a way to evaluate expressive language skills.Companies can select job appli-cants by this way and individuals can also improve themselves.However,most of the current evaluation methods for expression have respective shortcomings.For example,offline evaluation is time-consuming and PSC which is an online test has limitations because it focuses more on pronunciation.This thesis designs and implements an exam system for expressive language skills based on containers and microservices.The system not only has advantages of conve-nient and fast like online evaluation,but also improves the form of exam to meet more detailed evaluation needs.The system's evaluation form imitates the TOEFL oral test.An exam will consist of multiple question types,and different question types correspond to different directions of evaluation.Finally,the five dimensions of expressing ability,such as pronunciation,summarizing,detailing,structure and logic,will be scored sep-arately.In addition,the result of evaluation also includes a customized report,which will clearly state the current advantages and future improvement directions of the tested users.Besides the main exam interface,the system also provides a management inter-face for the administrator,through which the administrator can perform management operations for various items,including authorization,paper templates and questions in question base.To cope with future business growth,the system uses the containerized and mi-croservice architecture with good horizontal scalability.And the system client side is We Chat Mini Program,users can test anytime and anywhere because of this.The front-end and back-end of the system are separated.The back-end uses Python and provides API services in the form of Flask App,which will be called by the client.The mini program is built by Taro and Vue and Celery is used as a distributed task queue to decouple the two major processes,answering and judging.The algorithms of fea-ture extraction involved in judging questions are provided externally.The microservice module is consist of four parts: user service,exam service,analysis service and ques-tion service.Thrift RPC is used for communication between services.In addition,the corpus crawling module,cronjob module and celery worker module provide support for the operation of the system.Especially,all services are deployed using docker,and Kubernetes is used for management of containerized applications.The system can support 500 people to conduct exams simultaneously in the cluster deployment mode.Moreover,the mini program form reduces the learning cost of the system,and the sys-tem can be widely used in interviews and competitions to meet the needs of large-scale expression evaluation in a short period of time.Additionally,the system also provides a way for individual users to train their expressive language skills.