Research And Optimization Of Several Technologies In Online Examination System

With the rapid development of Internet technology, online examination system has been widely used in school examinations and personnel recruitment. The functions of automatic grading and generating score reports save a lot of manpower. Random-generating examination paper reduces the likelihood of plagiarism. At present, the function of online examination system has yet to be perfect, but there are still some disadvantages that question upload method is of low efficiency. In addition, online examination system is lack of front-end performance optimization solution, and when users make multiple concurrent requests, there will be disadvantages that result in page download time and time to start render too long. To solve these problems, this thesis designs and develops an online examination system with high performance.In this article, we first analyze question upload method in online examination system and introduce the basic principle of existing upload methods, analyzing their corresponding advantages and disadvantages in detail, and design a semantic-based, efficient and intelligent method named Intelligent Document Parsing(IDP). Then, we analyze factors that affect the performance of the Web front-end, putting forward the corresponding optimization solutions according to the factors, and apply the solutions to online examination system. Finally, we discuss the blocking problem caused by multiple concurrent requests under the condition of the limit number of simultaneous connections. This thesis proposes a request scheduling algorithm named APSC(Algorithm based on Priority queue for Simultaneous Connections). The specific work is as follows:(1) Existing methods of question upload need artificial pretreatment which makes them costly and inefficient. In order to solve this problem, a semantic-based efficient and intelligent method named IDP is proposed. Through rule matching and semantic analysis, IDP determines the type of the questions and obtains the correct information using discourse analysis method. IDP uses any form of the question documents directly, analyzing them intelligently and accurately, making full use of existing test resources, it has high generality and practicability.(2) Through the analysis on the HTTP protocol, the browser and the server interaction process, the browser cache mechanism and other factors that influence Web front-end, we propose an optimization solution for online examination system. The experimental results show that, after optimization, when first loaded, the average page download time is reduced by 18.8%, the optimization amplitude for time to start render is about 19.7%. When loading again, the average page download time is improved about 15.4%, time to start render is reduced by 9.2%.(3) In the case of limit number of simultaneous connections in a browser, making multiple requests at the same time will lead to the subsequent requests blocked, causing the page download time and time to start render too long. To solve this problem, we design a request scheduling algorithm named APSC. APSC considers the page download time, time to start render and request waiting time, not only ensures the page download time as little as possible, but also makes the page content presented to the user as soon as possible. In addition, APSC also avoids the phenomenon of the low priority requests waiting for a long time leading to "starvation". Experimental results show that, after optimization, page download time is reduced by 11% on average, time to start render is improved about 15%.