西安工商学院
毕业设计(论文)外文资料翻译
学 院
专 业
班 级
姓 名
学 号
外文出处
附 件1.原文；2.译文
2025年3月
Retracted: Design and Implementation of Online Japanese Examination System Based on Genetic Algorithm
In order to solve the problem of online inspection of students’ theoretical knowledge of Japanese, this paper further optimizes and adjusts the design of the online Japanese examination system and presents an online Japanese examination system based on genetic algorithm. Taking the Japanese test as the research object, on the basis of comprehensively analyzing the problems of slow test paper composition, low success rate, and low quality of traditional online test systems, an intelligent test composition model based on genetic algorithm is proposed, and the implementation process of genetic algorithm and the key steps are described in detail. The results show that the online Japanese examination system based on genetic algorithm can meet the needs of test paper generation in more complex situations. After a long time of operation and continuous improvement, the online Japanese examination system has obtained the adaptability of the best solution. The value of the fitness is 99.666667; when the fitness is at this value, the error of the question type score of the test paper is 0, the average difficulty error on the test paper is 0, and the error of the section test point distribution is 0.666667. This fully illustrates the stability and effectiveness of the Japanese online examination system, which can meet the needs of daily Japanese majors and improve the efficiency of Japanese teaching.
1. Introduction
Since the beginning of the new century, with the rapid development of science and technology, computer network technology has been gradually applied to all walks of life. At the same time, the continuous progress of computer network technology has brought many conveniences to many fields and provided technical support for the development of the education industry. For example, a large number of computers are used in the classroom and laboratory to deal with various tasks in the teaching process, which is more convenient, fast, and safe. Among them, we can use the intelligent computing advantages of genetic algorithms to develop and improve the online examination system. Based on the advantages and principles of genetic algorithm, this paper analyzes the advantages and disadvantages of the online examination system. At the same time, taking the Japanese examination as the research object, based on the comprehensive analysis of the problems of slow speed, low success rate, and low quality of the traditional online examination system, an intelligent paper generation model based on genetic algorithm is constructed to further improve and optimize the Japanese online examination system. Through an efficient and stable Japanese online examination system, we can effectively detect the problems existing in students' daily Japanese learning, improve the quality of Japanese teaching, and stimulate students' enthusiasm and initiative.
2. Related Works
Liu and others stated that China started late in the research of the online examination system. In 1998, China began to rise the online education platform and introduced the online examination system into the online education platform. The introduction of the online examination system improved the efficiency of examination management, reduced the work pressure of teaching staff, and solved the difficult problem of examinee’s remote examination [1]. Sugisawa and others stated that soon, major universities in China have successively developed online examination systems, among which the development of Shanghai Jiaotong University and Beijing University of Posts and Telecommunications is more prominent [2]. Alaqbi and others stated that after the twenty-first century, China’s science and technology has entered an era of rapid development [3]. In addition to colleges and universities, social training institutions have gradually introduced online examination systems, such as computer grade examination and driving school examination involving a wide range of people. Clivaz and others stated that with the progress of China’s science and technology, China’s online examination system technology has also made progress and effectively developed various online examination software to make the online examination system more comprehensive [4].
In recent years, Chinese experts have invested a lot of energy in the research of automatic online examination systems and have also made breakthroughs in this field and achieved fruitful results. The Dragon Online Examination System developed in recent years has a more complete functional system and can also be optimized for various test questions, ensuring that various operating interfaces are more convenient during the optimization process.
Mizuma and others stated that the first test developed algorithm in online testing was book test generation. Test paper generation is very inefficient, which makes it more efficient for teaching staff [5]. In addition, the quality of the papers varies, largely depending on the competence of the examiners. To address these issues, automated tests can be performed. Makino and others stated that these tests are not automatically computerized to select the required questions from the education test questions to conduct the test. The accuracy and quality of testing remains difficult to maintain. Therefore, experts have introduced technical testing on this basis, and the main research is the study of non-technical testing concepts automatic [6].
Lie and others stated that “intelligent test paper generation” has become the core technology of the online test system. For example, by adding complex intelligent algorithms to the automatic online examination system, the efficiency of online examination paper generation has been significantly improved [7]. Tian, Z, and others stated that the online examination system developed in recent years has become more and more complete, with a convenient and fast test paper generation function [8]. Acosta and others stated that the system is mainly applicable to enterprise-level examinations, has perfect functions and advantages, and can fundamentally solve users’ problems from reality [9]. For example, the operation steps are scientific and convenient, the test paper is convenient and fast, multiple people are supported online, and the confidentiality of the answer process is guaranteed. However, it has not developed more basic functions in the research of examination system technology, mainly expanding, reforming, and perfecting the original functions.
The analysis of domestic and foreign research by Cao et al. shows that the online examination system has a high reputation and is relatively developed [10]. At present, the research of the online examination system mainly focuses on the production of intelligent examination papers and the automatic scoring technology. With the technical support of genetic algorithm, intelligent technology can effectively improve the efficiency and accuracy of the online examination system, as shown in Figure 1.
3.Method
Genetic algorithms start with a potential population, which is a combination of multiple individuals with different codes. Chromosomes act as the main carrier, which determines the external shape of an individual. In the initial population, according to the principle of natural evolution, a better approximate solution is gradually generated. Individuals are selected according to their fitness, and then crossover and mutation are combined to generate a new population. This process enables the new population to be better than the initial population, and the optimal individual in the latest population can be used as an approximate optimal solution to the problem. The flowchart of the genetic algorithm is shown in Figure 2.
The genetic algorithm starts from the population and evaluates the individuals in the population, instead of searching from the individual. This is conducive to global selection, so the genetic algorithm is easier to achieve optimization. On the contrary, the traditional optimization algorithm searches for the individual, so it is extremely easy to fall into a local optimal solution. This is the advantage of the genetic algorithm, which differs from the traditional optimization algorithm. Genetic algorithms search based on probability, rather than performing deterministic orientation. Therefore, the search is larger, and the population is generated and the individuals in the population are evaluated. The genetic algorithm will organize the search by itself according to the fitness function and select individuals with large fitness to form a new group. So, it has strong organization and adaptability [11].
The initial population is composed of n individuals by using a random function. The first step is to measure the number of populations. A common approach is to record the population as 50 or multiples of 50 as the default. Past research has shown that population size is directly related to the success and quality of test paper production. If the population is unreasonably determined, the problem of local optima arises [12].
Coding method is the basis of the genetic algorithm. The level of coding directly determines the quality of problem-solving. Since the genetic algorithm was proposed, after years of development, many different coding methods have been formed. Among them, the most widely used are as follows: hybrid coding, binary coding, and real coding. In particular, genetic algorithms play a great role in the fields of function optimization, production scheduling, pattern recognition, neural networks, and adaptive control.
Figure 1: Design and implementation of online Japanese examination system based on genetic algorithm.
Figure 2: Flow chart of genetic algorithm.
The selection of the fitness function must meet two conditions: There is no “premature” phenomenon in the early stage, and there will be no “recession” in the later stage. Only the fitness function that meets these conditions can improve the fitness between individuals, reduce differentiation, and obtain the optimal results as a whole [13, 14]. In practice, this transformation process is relatively simple, as shown in the following formula:
When the total value of the desired optimization result is positive, it shows that the desired optimization result is basically consistent with the problem of individual fitness, as shown in the following formula:
Through the adjustment of the overall fitness function, objective function and constraints, the optimization between corresponding individuals is realized to ensure that the optimal solution can be obtained at present, as shown in the following formula:
Above, FðXÞ is the fitness function, α is the normal number, f(x) the objective function, and β is the constant coefficient. The objective function can be scaled and translated. There are many methods to determine the coefficient,as shown in the following formula (5):
Or as shown in the following formula (7):
The selection operator is based on the different fitness of different individuals. The state of its chromosomes in the next stage depends on the optimization degree and limit performance of the chromosomes. Individuals with high fitness will continue to be replicated in the next stage, while those with low fitness Individuals will be eliminated directly in the next stage. Its operation strategy is to retain the best individual, prevent the emergence of local optimal solution, and adhere to the elite retention strategy. The function of selecting operators in the genetic algorithm is to avoid the destruction of the Geshan gene and to improve the computational efficiency as a whole [15]. The main selection methods of the selection operator include the wheel selection method and the random selection method. This document mainly selects the selection method of the roulette, as shown in Figure 3.
The crossover operator is mainly calculated according to specific principles and methods. In the process of random selection of the whole population, it is necessary to ensure that the chromosomes meet the requirements of relevant exchange and groups were randomly paired and selected for crossover operation, which can make the overall optimization effect the best [16].
When using a genetic algorithm, certain control parameters must first be established in order for the algorithm to achieve its intended purpose. The control parameters include crossover rate, mutation rate, and termination iteration. In order to optimize the performance, the following parameter design methods are usually selected.
The operation process of the trial method is to first arrange and combine all the main parameters, then operate the parameters obtained by these combinations, and finally compare and comprehensively analyze the results obtained by the operation, and select the optimal control list.
The empirical method is based on the judgment of professional teachers or scholars and past research experience to determine the analysis value of specific parameters [3].
The above introduction shows in detail the application process and operation principle of genetic algorithm in intelligent test paper generation. It also simply points out the problems and avoidance methods of genetic algorithm in the survival process of the “chromosome” fittest. To improve operational efficiency and optimize results, it acts as a smart module during the special process of making the test paper.
Figure 3: Schematic diagram of wheel disc selection
With various combinations of variants and intersections, designs can be considered in a repetitive way of thinking so that the individual situation can be played from the population, so as to ensure that the whole variation is only controlled within the mother for observation. The operation flow chart of the test paper generation system is shown in Figure 4.
In the overall operation, each participating functional module will be coded to facilitate the real-time search of constraint groups during actual operation. The design of coding should be distinguished according to different question types. Class structure design mainly designs the main module classes and the relationship between classes and uses the way of class diagram to express. According to the question types, a fixed corresponding function module can be directly established. An example of designing a partition code is shown in Table 1.
In addition to encoding a large number of individuals with question numbers, there are M questions in the whole question bank, which need to be distinguished by binary string description documents, as shown in the following formula:
If the total score of the test paper is set to Total Mark, the following calculation formula can be obtained, as shown in the following formula:
Figure 4: Operation flow chart of test paper generation system.
Table 1: Example of designing a partition code.
In this formula, set the frequency of each number in the weighted average of each component. The formula is shown in the following formula:
When using a special algorithm, the problem is solved in terms of the minimum value of the motion function so that the corresponding motion function is included as follows:
The basic requirements of fitness function are as follows: One is to ensure that all variables are positive in the case of input. The second is to ensure consistent progress throughout the optimization process. The fitness function can construct various benefit parameters, and the fitness function can also be used to represent various individual spaces (S). The corresponding function (F) is shown as follows:
The three elements constituting the total fitness function are summed to obtain the value of the total fitness function. The formula is shown in the following formula:
After long-term hard work and continuous improvement of Japanese online testing, the genetic algorithm has gained the strength of the best solution after long-term development and growth. The physical value is 99.666667. When this value is output, the error of the test result query type is 0, the error of the average difficulty of the test is 0, and the error of the distribution of some test points is 0.666667. The process of obtaining the fitness of this optimal solution is shown in Figure 5.
(1) Basic knowledge of Japanese:
It refers to the basic knowledge and skills that students must master in Japanese learning.These Japanese knowledge and skills are tested in the test paper.
Figure 5: Fitness curve of optimal solution.
(2) Japanese examination outline. Reflect the main direction of the Japanese test to the examinee, and prompt the key knowledge points
(3) The difficulty of Japanese test. It mainly refers to the difficulty and ease of the examinee to correctly answer the test question. The difficulty of objective questions can be calculated by the following formula, as shown in the following formula:
The complexity of the subjective test questions can be calculated using the following formula, as shown in the following formula:
Combined with the actual needs, we can set up a set of mathematical model, and then set some constraint variables for this mathematical model. Controlling the complexity of the overall question types on the Japanese electronic test paper, it is necessary to set up a set of standard test paper mechanism. The NP complete problem is a subclass of the NP class problems, a subclass with special properties and special meaning. Before generating the test paper, we should first determine the question type distribution, test site range, difficulty coefficient, score distribution, etc., of the whole set of test paper. Combined with the actual situation, we can ensure the scientificity and reference value of test paper generation, as shown in Tables 2–3.
Table 2: Information of objective questions such as single choice questions.
4.Experiment and Analysis
Establish an examination database named "in" and complete the creation of the required data table. After the database is created, the connection between the foreground interface of the system and the background database needs to be established. The connection operation to the database is used repeatedly in the program. Therefore, the functions of database connection, query object creation, and result set creation are encapsulated in a name. The database connection of the system adopts data connection pool technology.
As a detection means, the examination should be serious and have special high requirements for security. Therefore, users are divided into two categories: administrators and candidates [17]. They have different levels. When entering the system, they need authentication and need to enter the authentication code. The administrator account is specifically set up for people who can make whole-system changes to the computer, install programs, and access all the files on the computer. Only users who have the administrator have full access to other user accounts on the computer. The authentication identification code here is the unique random number directly generated by the system for each server connection. Users must enter the identification code when logging in; otherwise, they will not be able to log in. Especially in the student Japanese examination, once the candidate opens the browser and enters the login interface, the corresponding unique ID code will be generated. At this time, the system will automatically track the user. Before the end of the examination, the user is not allowed to log in again, so the candidate cannot log in from multiple windows and places at the same time, so as to effectively avoid the students cheated in the Japanese online exam, as shown in Figure 6.
After passing the authentication, students enter the online examination home page, select the specific content of the Japanese test, and call out the test paper for examination. The candidates’ answers are saved on the server in the form of documents. During the Japanese online examination, you can click to submit the answer after completing each page of the examination questions and then do the examination questions on the next page. If you want to quit the exam halfway, you can click "quit halfway." The system will automatically display the test questions and correct answers that you have taken before you quit and make a comparison to get the score. If all the test questions are completed or the test time is over, the system will display all the test questions and their standard answers and compare them with the answers made by the user and give scores.
The test question upload interface can use the function of batch upload in the process of uploading test questions, which can effectively reduce the burden of the database [18, 19]. Testing the question bank adjustment interface in the test question information management module, the administrator can view, delete, and modify the content of the test question bank.
After using the online Japanese examination system, the whole examination process is as follows:The examinee enters the entrance page of the online Japanese examination system through the browser.After entering the correct user information, the examinee enters the Japanese examination system and randomly selects the test questions.After clicking "start," the system will automatically generate the test paper and display it on the client through the browser.The system starts timing and begins the exam.After the examinee submits the test paper, the computer automatically marks the objective questions, obtains the score, and counts the score into the database.Subjective questions such as short answer questions can be marked by a combination of computer and manual.
Therefore, the online Japanese examination system should consider the following functions:Realize the management of the examination question bank, such as the establishment of the question bank, the entry, modification, and deletion of questions. The types of questions in the question bank are mainly objective questions such as single choice, blank filling, and judgment.Students log in to the system on the Internet and randomly select test questions for examination. During the examination, it can automatically count the time and display the students’ remaining time. At the end of the examination, it can automatically take up the paper and change the paper [20].
Students’ test scores can be queried online, and their own test papers can be rechecked.Realize the seamless link with the college educational administration management system, and students’ scores can be automatically imported into the educational administration management system to facilitate students’ query.Japanese teachers can set the test parameters before the test, such as test subjects, test time, and test scope.After the examination, the Japanese teacher can analyze the results of the examination, such as the score of each chapter and question type, and the score statistics of each class.Users logging into the system, including system administrators, teachers, and students, should have different permissions [21].
Table 3: Information table of subjective questions such as question and answer questions
Figure 6: System login interface
Log in to different function interfaces according to permissions.The process of administrator login management module is shown in Figure 7.The administrator login management module function is used to verify the administrator’s identity when entering the examination management system. When logging in the examination management system, the system allows the administrator to express his identity. Different administrators have different permissions. The system verifies whether the administrator is a legal user. When the user name and password entered by the administrator are correct, he can enter the management system. Open corresponding management functions for administrators according to different permissions. A complete software testing management tool should be able to manage all aspects of the testing process.The flow of teacher login management module is shown in Figure 8.
The function of the teacher login management module is mainly to complete the test paper generation, and submission after the teacher is authorized. The system stores the test paper in the database test paper table to generate the form of question bank for students to extract the test and view the results. Teachers can also query student users, enter scores, and publish examination arrangements and other news information. The teacher can review the students’ homework.The process of student user login examination module is shown in Figure 9.
Figure 7: System flow chart of administrator login management function.
Figure 8: Teacher login management module process.
Student user authentication means that during the Japanese online exam, the examinee can log in to the system to take the exam only after he/she has passed face recognition or ID card recognition [23]. Identity authentication is very important in the process of Japanese online examination. It is related to the seriousness of the examination.
The online Japanese examination system supports the exemption of the invigilation link, and all the examination room requirements can be achieved through the examination setting. Therefore, the main thing in the Japanese examination system is the various functions of the examination room environment, such as the test time, the list of candidates, the topic selection group papers, and the scoring criteria. After logging in, students can choose subjects and take tests. After the exam, they can see the correct answers and scores of the questions they have done.
The general input and output stream classes in Java language use the single byte reading method for data I/O operation. That is, only one byte of data is read or written at a time. This method is obviously cumbersome and inefficient. The reading and writing process is shown in Figure 10.
Because the Java language provides a buffer class specially used to improve the I/O efficiency of the system, it is like providing a temporary buffer when reading and writing data. A data block of buffer size can be read at a time. It can reduce the number of read and write data. If you transmit only a little data each time, you need to be transmitted many times, which will waste a lot of time. And then this data block can be written to the target device at one time [24]. Setting up a data buffer to read one data block at a time to improve system performance is particularly important in network data transmission. The flow chart of reading and writing using data buffer class is shown in Figure 11.
For the verification of the existence and legitimacy of the input information from the client, it is realized by client-based programming. For example, the student and teacher administrators must pass the verification of the existence and legitimacy of the input from the client when registering or logging in. Specifically, if the information that must be filled in is not filled in, the system will give an existential warning. For example, for those systems that do not meet the requirements in terms of type composition, length extreme value, and so on, the system will also give a warning on the client. This can greatly reduce the burden of the server and improve the reliability of the program and the running speed of the system.
Figure 9: Student user login examination module process.
Figure 10: Single byte read and write.
Figure 11: Reading and writing process using data buffer class.
Aiming at server-side user exceptions in the server-side database, the server-side programming is used to correct the user’s misbehavior to ensure the normal operation of the system. For example, when the data content, data type, data format, and data scheduled processing process are obtained by the system from the client conflict with the database in the server, the system can give a warning in time and guide the correct processing method. In this system, this abnormal message processing mechanism runs through the processing of all data and shows good fault-tolerant performance in the practice of remote test run of the Japanese examination system [25].
And the system has very strict identity authentication procedures. Student administrators must be authenticated to log in, and their permissions are different for different users. For example, different administrators can only operate with their own permissions. Through strict identity authentication and different authority settings, the security of the online examination system and data security are guaranteed.
In addition, we also optimize the source code by writing modular functions and encapsulating process code in an object-oriented way to improve the reusability and execution efficiency of the source program.
In order to complete the test of all the learning contents of the Japanese course, each database has a data table according to the type of questions (single-choice, fill-in-the-blank, true-false, and short-answer questions) [26]. When the administrator sets the test parameters, he can select the test subjects. The following tables are established in the question bank: multiple-choice question sheet, fill-in-the-blank question sheet, judgment question sheet, and short-answer question sheet, which are used to store questions of various types.
Add difficulty, bias, and other attributes to each question to improve test quality. The student information table is used to store important student information, usually including student number, ID number, name, department chair, and other items. Information can be sent through the University’s Academic Administration. The user table is used to store management information, including number, user name, password, authorization, and other equipment. The teacher table is used to store teacher information, including account number, password, name, office, and other items. The quiz table is used to store quizzes created by the quiz algorithm. Answers are used to store answers to student tests. We can also establish a score table to make statistics and analysis of students’ scores.
In this way, the online Japanese examination system can be realized both functionally and safely, which improves the convenience of Japanese examination.
5. Conclusion
The genetic algorithm is based on the theory of evolution and can complete the intelligent questionnaire very well. In this paper, a mathematical model of the genetic algorithm is established for the test-setting part of the online Japanese examination system, and then the model is realized by editing the code. Genetic algorithm can solve many deficiencies in the traditional way of setting test papers, such as low efficiency and poor quality of test papers, which greatly optimizes the fairness and poor reference of traditional test papers. After a period of testing and analysis of the system, the functional modules of the online Japanese examination system can stably carry out various tasks such as grouping papers, but there are still many imperfections, and various drawbacks are still unavoidable. There are still many undiscovered problems, and various problems must be gradually discovered in the long-term operation and use. By constantly finding problems and solving problems, the system can be made more perfect and more practical.
References
[1] Z. Liu, J. Liu, and Z. Liu, “Analysis, design, and implementation of impulse-injection-based online grid impedance identification with grid-tied converters,” IEEE Transactions on Power Electronics, vol. 35, no. 12, pp. 12959–12976, 2020.
[2] H. Sugisawa, T. Shinoda, Y. Shimizu, and T. Kumagai, “Cognition and implementation of disaster preparedness among Japanese dialysis facilities,” International Journal of Nephrology, vol. 2021, Article ID 6691350, 9 pages, 2021.
[3] A. Al-Aqbi, R. Al-Taie, and S. K. Ibrahim, “Design and implementation of online examination system based on msvs and SQL for university students in Iraq,” Webology, vol. 18, no. 1, pp. 416–430, 2021.
[4] S. Clivaz and T. Miyakawa, “The effects of culture on mathematics lessons: an international comparative study of a collaboratively designed lesson,” Educational Studies in Mathematics, vol. 105, no. 1, pp. 53–70, 2020.
[5] M. Mizuma, H. Yamamoto, H. Miyata et al., “Impact of a board certification system and implementation of clinical practice guidelines for pancreatic cancer on mortality of pancreaticoduodenectomy,” Surgery Today, vol. 50, no. 10, pp. 1297–1307, 2020.
[6] S. Makino and D. Lehmberg, “The past and future contributions of research on Japanese management,” Asian Business & Management, vol. 19, no. 1, pp. 1–7, 2020.
[7] Z. W. Lie, Q. L. Zheng, S. Zhou, and H. L. Rauf, “Virtual energy-saving environmental protection building design and implementation,” International Journal of System Assurance Engineering and Management, vol. 13, Supplement 1, pp. 263–272, 2022.
[8] Z. Tian, S. Tian, T. Wang, Z. Gong, and Z. Jiang, “Design and implementation of open source online evaluation system based on cloud platform,” Journal on Big Data, vol. 2, no. 3, pp. 117–123, 2020.
[9] J. Acosta, F. Amórtegui, A. Escobar, L. M. Leon, and S. Rivera, “Design and implementation of prototype for XLPE cable aging test,” Revista Internacional de Métodos Numéricos para Cálculo y Diseño en Ingeniería, vol. 36, no. 3, pp. 36–44, 2020.
[10] M. Cao, “Design and implementation of multidimensional interaction in online English course under the assistance of Omnimedia,” Scientific Programming, vol. 2021, Article ID 3713161, 10 pages, 2021.
[11] N. Choudhary, “Design and implementation of wildfire monitoring system,” International Journal for Modern Trends in Science and Technology, vol. 7, no. 5, pp. 139–143, 2021.
[12] K. Sharma and B. K. Chaurasia, “Trust based location finding mechanism in VANET using DST,” in Fifth International Conference on Communication Systems & Network Technologies, pp. 763–766, IEEE, Gwalior, India, 2015.
[13] A. Muneer and D. Zhan, “Design and implementation of automatic painting mobile robot,” IAES International Journal of Robotics and Automation (IJRA), vol. 10, no. 1, pp. 68–74, 2021.
[14] P. Elechi and C. O. Ahiakwo, “Design and implementation of an automated security gate system using global system for mobile communication network,” Journal of Network and Computer Applications, vol. 7, no. 1, pp. 1–10, 2021.
[15] S. Kaddoura, D. E. Popescu, and J. D. Hemanth, “A systematic review on machine learning models for online learning and examination systems,” PeerJ Computer Science, vol. 8, no. e986, 2022.
[16] L. Hu, “Design and implementation of a component-based intelligent clothing style CAD system,” Computer-Aided Design and Applications, vol. 18, no. S1, pp. 22–32, 2020.
[17] J. Li and W. Li, “On-line PID parameters optimization control for wind power generation system based on genetic algorithm,” IEEE Access, vol. 8, pp. 137094–137100, 2020.
[18] R. Dutta, A. Mantri, and G. Singh, “Evaluating system usability of mobile augmented reality application for teaching Karnaugh-maps,” Smart Learning Environments, vol. 9, no. 1, p. 6, 2022.
[19] B. Alhijawi, Y. Kilani, and A. Alsarhan, “Improving recommendation quality and performance of genetic-based recommender system,” International Journal of Advanced Intelligence Paradigms, vol. 15, no. 1, pp. 77–88, 2020.
[20] S. Ghareeb, A. J. Hussain, D. Al-Jumeily et al., “Evaluating student levelling based on machine learning model’s performance,” Discover Internet of Things, vol. 2, no. 1, p. 3, 2022.
[21] B. Xia, X. Zheng, L. Zhang, and L. Zhao, “UWB positioning system based on genetic algorithm,” Journal of Computer and Communications, vol. 9, no. 4, pp. 110–118, 2021.
[22] M. A. Elaziz, A. A. Ewees, and Z. Alameer, “Improving adaptive neuro-fuzzy inference system based on a modified salp swarm algorithm using genetic algorithm to forecast crude oil price,” Natural Resources Research, vol. 29, no. 4, pp. 2671–2686, 2020.
[23] A. Gholami-Rahimabadi, H. Razmi, and H. Doagou-Mojarrad, “Multiple-deme parallel genetic algorithm based on modular neural network for effective load shedding,” Soft Computing, vol. 25, no. 21, pp. 13779–13794, 2021.
[24] Z. Xu, M. M. Kamruzzaman, and J. Shi, “Method of generating face image based on text description of generating adversarial network,” Journal of Electronic Imaging, vol. 31, no. 5, article 051411, 2022.
[25] W. Xia and L. Shen, “Joint resource allocation at edge cloud based on ant colony optimization and genetic algorithm,” Wireless Personal Communications, vol. 117, no. 2, pp. 355–386, 2021.
[26] P. Rani, R. Kumar, A. Jain, and S. K. Chawla, “A hybrid approach for feature selection based on genetic algorithm and recursive feature elimination,” International Journal of Information System Modeling and Design, vol. 12, no. 2, pp. 17–38, 2021.
在线日语考试的设计与实现-基于遗传算法的系统
为了解决学生日语理论知识在线考试的问题，本文进一步优化和调整了在线日语考试系统的设计，提出了一种基于遗传算法的在线日语考试体系。以日语考试为研究对象，在综合分析传统在线考试系统组卷速度慢、成功率低、质量低等问题的基础上，提出了一种基于遗传算法的智能组卷模型，详细描述了遗传算法的实现过程和关键步骤。结果表明，基于遗传算法的在线日语考试系统可以满足更复杂情况下的试卷生成需求。经过长时间的运行和不断改进，在线日语考试系统已经获得了最佳解决方案的适应性。适合度值为99.666667；当适合度为该值时，试卷题型得分的误差为0，试卷上的平均难度误差为0。分段测试点分布的误差为0.666667。这充分说明了日语在线考试系统的稳定性和有效性，可以满足日常日语专业学生的需求，提高日语教学效率。
1.导言
进入新世纪以来，随着科学技术的快速发展，计算机网络技术逐渐应用于各行各业。与此同时，计算机网络技术的不断进步给许多领域带来了许多便利，为教育产业的发展提供了技术支持。例如，在课堂和实验室中使用大量的计算机来处理教学过程中的各种任务，这更方便、快速、安全。其中，我们可以利用遗传算法的智能计算优势来开发和改进在线考试系统。本文基于遗传算法的优点和原理，分析了在线考试系统的优缺点。同时，以日语考试为研究对象，在综合分析传统在线考试系统速度慢、成功率低、质量低等问题的基础上，构建了基于遗传算法的智能组卷模型，以进一步改进和优化日语在线考试系统。通过高效稳定的日语在线考试系统，我们可以有效地发现学生日常日语学习中存在的问题，提高日语教学质量，激发学生的积极性和主动性。
2.相关工作
刘等人表示，中国在线考试系统的研究起步较晚。1998年，中国开始兴起在线教育平台，并将在线考试系统引入在线教育平台。在线考试系统的引入提高了考试管理的效率，减轻了教职员工的工作压力，解决了考生远程考试的难题[1]。Sugisawa等人表示，不久之后，中国各大高校相继开发了在线考试系统，其中上海交通大学和北京邮电大学的发展更为突出[2]。阿拉克比等人表示，21世纪后，中国的科学技术进入了一个快速发展的时代[3]。除了高校，社会培训机构也逐步引入了在线考试系统，如计算机等级考试和驾驶学校考试，涉及范围广泛。Clivaz等人表示，随着中国科技的进步，中国的在线考试系统技术也取得了进步，有效开发了各种在线考试软件，使在线考试系统更加全面[4]。
近年来，中国专家在自动在线考试系统的研究上投入了大量精力，并在该领域取得了突破，取得了丰硕成果。近年来开发的龙在线考试系统功能系统更加完善，还可以针对各种试题进行优化，确保在优化过程中各种操作界面更加方便。
Mizuma和其他人表示，在线测试中第一个测试开发的算法是书籍测试生成。试卷生成效率非常低，这使得教师更有效率[5]。此外，论文的质量各不相同，很大程度上取决于考官的能力。为了解决这些问题，可以进行自动化测试。Makino和其他人表示，这些测试不是自动计算机化的，无法从教育测试问题中选择所需的问题来进行测试。测试的准确性和质量仍然难以保持。因此，专家们在此基础上引入了技术测试，主要研究的是非技术测试概念的自动化[6]。
李等人指出，“智能组卷”已成为在线考试系统的核心技术。例如，通过在自动在线考试系统中添加复杂的智能算法，在线试卷生成的效率得到了显著提高[7]。田、Z等人表示，近年来开发的在线考试系统越来越完善，具有方便快捷的试卷生成功能[8]。Acosta等人表示，该系统主要适用于企业级考试，功能完善，优势明显，能够从根本上解决用户的实际问题[9]。例如，操作步骤科学方便，试卷方便快捷，支持多人在线，保证了答案过程的保密性。然而，在考试系统技术的研究中，它还没有开发出更多的基本功能，主要是对原有功能的扩展、改革和完善。
曹等人对国内外研究的分析表明，在线考试系统具有较高的声誉，相对发达[10]。目前，在线考试系统的研究主要集中在智能试卷的生成和自动评分技术上。在遗传算法的技术支持下，智能技术可以有效提高在线考试系统的效率和准确性，如图1所示。
3.方法
遗传算法从一个潜在的群体开始，这个群体是具有不同编码的多个个体的组合。染色体是决定个体外形的主要载体。在初始种群中，根据自然进化原理，逐渐生成更好的近似解。根据个体的适应度进行选择，然后将交叉和突变结合起来，生成一个新的群体。这个过程使新的种群比初始种群更好，并且最新种群中的最优个体可以用作问题的近似最优解。遗传算法的流程图如图2所示。
遗传算法从群体开始，评估群体中的个体，而不是从个体开始搜索。这有利于全局选择，因此遗传算法更容易实现优化。相反，传统的优化算法会搜索个体，因此很容易陷入局部最优解。这就是遗传算法不同于传统优化算法的优点。遗传算法基于概率进行搜索，而不是执行确定性定向。因此，搜索范围更大，生成群体并评估群体中的个体。遗传算法将根据适应度函数自行组织搜索，并选择适应度较大的个体组成一个新的群体。因此，它具有很强的组织性和适应性[11]。
通过使用随机函数，初始种群由n个个体组成。第一步是测量人口数量。一种常见的方法是将人口记录为50或50的倍数作为默认值。过去的研究表明，人口规模与试卷制作的成功和质量直接相关。如果人口被不合理地确定，则会出现局部最优问题[12]。
编码方法是遗传算法的基础。编码水平直接决定了问题解决的质量。自遗传算法提出以来，经过多年的发展，已经形成了许多不同的编码方法。其中，应用最广泛的有：混合编码、二进制编码和实数编码。特别是，遗传算法在函数优化、生产调度、模式识别、神经网络和自适应控制等领域发挥着重要作用。
图1：基于遗传算法的在线日语考试系统的设计与实现
图2：遗传算法的流程图
适应度函数的选择必须满足两个条件：早期没有“过早”现象，后期不会有“衰退”。只有满足这些条件的适应度函数才能提高个体之间的适应度，减少分化，并从整体上获得最佳结果[13,14]。在实践中，这种转换过程相对简单，如下式所示：
当期望优化结果的总值为正时，表明期望优化结果与个体适应性问题基本一致，如下式所示：
通过调整整体匹配函数、目标函数和约束条件，实现相应个体之间的优化，以确保目前可以获得最优解，如下式所示：
上面，fx)是适配函数，α是正态数，F（X）是目标函数，β是常数系数。目标函数可以缩放和转换。确定系数的方法有很多，如下式（5）所示：
或如下式（7）所示：
选择算子基于不同个体的不同适应度。其染色体在下一阶段的状态取决于染色体的优化程度和极限性能。下一阶段将继续复制健康状况良好的个体，而健康状况较差的个体将在下一阶段直接被淘汰。其运营策略是留住最优秀的人才，防止出现局部最优解，坚持精英留住策略。遗传算法中选择算子的功能是避免格山基因的破坏，提高整体计算效率[15]。选择算子的主要选择方法包括轮选择法和随机选择法。本文档主要选择轮盘赌的选择方法，如图3所示。
交叉算子主要依据特定的原理和方法进行计算。在对整个群体进行随机选择的过程中，必须确保染色体满足相关交换的要求，并随机配对和选择组进行交叉操作，这样可以使整体优化效果达到最佳[16]。
当使用遗传算法时，必须首先建立某些控制参数，以便算法实现其预期目的。控制参数包括交叉率、变异率和终止迭代。为了优化性能，通常会选择以下参数设计方法。
试验法的操作过程是首先对所有主要参数进行整理和组合，然后对这些组合得到的参数进行操作，最后对操作得到的结果进行比较和综合分析，选择最优控制列表。
实证方法是基于专业教师或学者的判断和以往的研究经验来确定具体参数的分析值[3]。
以上介绍详细介绍了遗传算法在智能组卷中的应用过程和工作原理。并简单指出了遗传算法在“染色体”适者生存过程中存在的问题和避免方法。为了提高操作效率并优化结果，它在制作试卷的特殊过程中充当智能模块。
图3：轮盘选择示意图
通过变体和交叉点的各种组合，可以以重复的思维方式考虑设计，以便从人群中发挥个体情况，从而确保整个变体仅在母体内控制以供观察。试卷生成系统的操作流程图如图4所示。
在整体操作中，每个参与的功能模块都将被编码，以方便在实际操作中实时搜索约束组。编码的设计应根据不同的问题类型进行区分。类结构设计主要设计主要模块类和类之间的关系，并采用类图的方式进行表达。根据问题类型，可以直接建立固定的对应功能模块。表1中示出了设计分区代码的示例。
除了用问题编号对大量个体进行编码外，整个题库中还有M个问题，需要用二进制字符串描述文档进行区分，如下公式所示：
如果试卷的总分设置为总分，则可以得到以下计算公式，如以下公式所示：
图4：试卷生成系统的操作流程图
表1：设计分区代码的示例
在此公式中，设置每个分量的加权平均值中每个数字的频率。公式如下：
当使用特殊算法时，根据运动函数的最小值来解决问题，因此相应的运动函数如下：
适配函数的基本要求如下：一是确保在输入的情况下所有变量都是正的。第二，确保在整个优化过程中取得一致的进展。适合度函数可以构造各种有益参数，适合度函数也可以用来表示各种单独的空间（S）。相应的函数（F）如下：
将构成总匹配度函数的三个元素相加，得到总匹配度的值。公式如下：
经过长期的努力和日本在线测试的不断改进，遗传算法经过长期的发展和壮大，已经获得了最佳解决方案的实力。物理值为99.666667。输出此值时，测试结果查询类型的误差为0，测试平均难度的误差为零，部分测试点的分布误差为0.666667。图5显示了获得该最优解的适应度的过程。
（1） 日语基础知识：
它是指学生在日语学习中必须掌握的基本知识和技能。这些日语知识和技能在试卷中进行了测试。
图5：最优解的适应度曲线
（2） 日语考试大纲。向考生反映日语考试的主要方向，并提示关键知识点
（3） 日语考试的困难。它主要指考生正确回答试题的难易程度。客观问题的难度可以通过以下公式计算，如下式所示：
主观试题的复杂性可以用以下公式计算，如下公式所示：
结合实际需要，我们可以建立一套数学模型，然后为这个数学模型设置一些约束变量。为了控制日本电子试卷整体问题类型的复杂性，有必要建立一套标准的试卷机制。NP完全问题是NP类问题的一个子类，具有特殊性质和特殊意义。在生成试卷之前，我们应该首先确定整套试卷的问题类型分布、考点范围、难度系数、分数分布等。结合实际情况，我们可以确保试卷生成的科学性和参考价值，如表2-3所示。.
表2：单选题等客观题信息
4.实验与分析
建立一个名为“in”的考试数据库，并完成所需数据表的创建。创建数据库后，需要建立系统前台接口和后台数据库之间的连接。与数据库的连接操作在程序中重复使用。因此，数据库连接、查询对象创建和结果集创建的功能都封装在一个名称中。系统的数据库连接采用数据连接池技术。
作为一种检测手段，检查应该是严肃的，对安全性有特殊的高要求。因此，用户分为两类：管理员和候选人[17]。它们有不同的层次。进入系统时，他们需要身份验证，需要输入身份验证码。管理员帐户是专门为可以对计算机进行整个系统更改、安装程序和访问计算机上所有文件的人设置的。只有具有管理员权限的用户才能完全访问计算机上的其他用户帐户。这里的身份验证标识码是系统为每个服务器连接直接生成的唯一随机数。用户登录时必须输入识别码；否则，他们将无法登录。特别是在学生日语考试中，一旦考生打开浏览器并进入登录界面，就会生成相应的唯一ID码。此时，系统将自动跟踪用户。考试结束前，用户不允许再次登录，因此考生不能同时从多个窗口和多个地方登录，从而有效避免了学生在日语在线考试中作弊，如图6所示。
通过认证后，学生进入在线考试主页，选择日语考试的具体内容，并调出试卷进行考试。考生的答案以文档的形式保存在服务器上。在日语在线考试中，您可以在完成每一页试题后点击提交答案，然后在下一页完成试题。如果你想中途退出考试，你可以点击“中途退出”。系统会自动显示你退出前的试题和正确答案，并进行比较以获得分数。如果所有试题都已完成或考试时间结束，系统将显示所有试题及其标准答案，并将其与用户的答案进行比较并给出分数。
试题上传界面可以在上传试题的过程中使用批量上传功能，可以有效减轻数据库的负担[18,19]。在试题信息管理模块中测试题库调整界面，管理员可以查看、删除和修改题库的内容。
使用在线日语考试系统后，整个考试过程如下：考生通过浏览器进入在线日语考试的入口页面。输入正确的用户信息后，考生进入日语考试系统，随机选择试题。点击“开始”后，系统会自动生成试卷，并通过浏览器在客户端显示。系统开始计时并开始考试。考生提交试卷后，计算机会自动标记客观问题，获得分数，并将分数计入数据库。主观问题，如简答题，可以通过计算机和人工相结合的方式进行标记。
因此，在线日语考试系统应考虑以下功能：实现题库的管理，如题库的建立、试题的录入、修改和删除。题库中的问题类型主要是客观问题，如单选、填空和判断。学生在互联网上登录系统，随机选择试题进行考试。在考试过程中，它可以自动计算时间并显示学生的剩余时间。考试结束时，它可以自动拿起试卷并更换试卷[20]。
学生的考试成绩可以在线查询，他们自己的试卷可以重新检查。实现与高校教务管理系统的无缝链接，学生成绩可以自动导入教务管理系统，方便学生查询。日语教师可以在考试前设置考试参数，如考试科目、考试时间和考试范围。考试结束后，日语老师可以分析考试结果，如每章和问题类型的分数，以及每节课的分数统计。登录系统的用户，包括系统管理员、教师和学生，应具有不同的权限[21]。
表3：问答题等主观题信息表
图6：系统登录界面
根据权限登录不同的功能界面。管理员登录管理模块的流程如图7所示。管理员登录管理功能用于在进入考试管理系统时验证管理员的身份。登录考试管理系统时，系统允许管理员表达自己的身份。不同的管理员具有不同的权限。系统验证管理员是否为合法用户。当管理员输入的用户名和密码正确时，他可以进入管理系统。根据不同权限为管理员打开相应的管理功能。一个完整的软件测试管理工具应该能够管理测试过程的各个方面。教师登录管理模块的流程如图8所示。
教师登录管理模块的功能主要是完成试卷生成，并在教师授权后提交。系统将试卷存储在数据库试卷表中，生成题库形式，供学生提取试卷并查看结果。教师还可以查询学生用户，输入分数，发布考试安排和其他新闻信息。教师可以查看学生的作业。学生用户登录考试模块的过程如图9所示。
图7：管理员登录管理功能的系统流程图.
图8：教师登录管理模块流程
学生用户认证意味着在日语在线考试期间，考生只有在通过人脸识别或身份证识别后才能登录系统参加考试[23]。身份认证在日语在线考试过程中非常重要。这与考试的严肃性有关。
在线日语考试系统支持免除监考环节，所有考场要求都可以通过考试设置来实现。因此，日语考试系统的主要功能是考场环境的各种功能，如考试时间、考生名单、选题组试卷和评分标准。登录后，学生可以选择科目并参加考试。考试结束后，他们可以看到他们所做问题的正确答案和分数。
Java语言中的通用输入和输出流类使用单字节读取方法进行数据I/O操作。也就是说，一次只能读取或写入一个字节的数据。这种方法显然既繁琐又低效。读写过程如图10所示。
因为Java语言提供了一个专门用于提高系统I/O效率的缓冲区类，所以它就像在读写数据时提供一个临时缓冲区。一次可以读取缓冲区大小的数据块。它可以减少读写数据的数量。如果每次只传输少量数据，则需要多次传输，这将浪费大量时间。然后，该数据块可以一次写入目标设备[24]。在网络数据传输中，设置数据缓冲区一次读取一个数据块以提高系统性能尤为重要。使用数据缓冲类进行读写的流程图如图11所示。
为了验证客户端输入信息的存在性和合法性，通过基于客户端的编程来实现。例如，学生和教师管理员在注册或登录时必须通过客户端输入的存在性和合法性的验证。具体来说，如果必须填写的信息未填写，系统将发出存在警告。例如，对于那些在类型组成、长度极值等方面不符合要求的系统，系统也会向客户端发出警告。这可以大大减轻服务器的负担，提高程序的可靠性和系统的运行速度。
图9：学生用户登录考试模块流程
图10：单字节读写
图11：使用数据提供类的读写过程.
针对服务器端数据库中的服务器端用户异常，使用服务器端编程来纠正用户的不当行为，以确保系统的正常运行。例如，当系统从客户端获取的数据内容、数据类型、数据格式和数据调度处理过程与服务器中的数据库冲突时，系统可以及时发出警告并指导正确的处理方法。在该系统中，这种异常消息处理机制贯穿于所有数据的处理过程，并在日语考试系统的远程试运行实践中表现出良好的容错性能[25]。
该系统具有非常严格的身份验证程序。学生管理员必须经过身份验证才能登录，不同用户的权限不同。例如，不同的管理员只能使用自己的权限进行操作。通过严格的身份认证和不同的权限设置，保证了在线考试系统的安全性和数据安全。
此外，我们还通过编写模块化函数和以面向对象的方式封装过程代码来优化源代码，以提高源程序的可重用性和执行效率。
为了完成日语课程所有学习内容的测试，每个数据库都有一个根据问题类型（单选、填空、真假和简答题）的数据表[26]。管理员设置测试参数时，可以选择测试对象。题库中建立了以下表格：多项选择题、填空题、判断题和简答题，用于存储各种类型的问题。
为每个问题添加难度、偏见和其他属性，以提高测试质量。学生信息表用于存储重要的学生信息，通常包括学生编号、身份证号码、姓名、系主任等项目。信息可以通过大学的学术管理部门发送。用户表用于存储管理信息，包括编号、用户名、密码、授权和其他设备。教师表用于存储教师信息，包括帐号、密码、姓名、办公室和其他项目。测验表用于存储由测验算法创建的测验。答案用于存储学生测试的答案。我们还可以建立一个成绩表，对学生的成绩进行统计和分析。
通过这种方式，在线日语考试系统可以在功能上和安全上实现，提高了日语考试的便利性。
5.结论
遗传算法基于进化论，可以很好地完成智能问卷。本文为在线日语考试系统的考试设置部分建立了遗传算法的数学模型，然后通过编辑代码实现了该模型。遗传算法可以解决传统试卷设置方式中的许多不足，如效率低、试卷质量差，大大优化了传统试卷的公平性和参考性差。经过一段时间的系统测试和分析，在线日语考试系统的功能模块可以稳定地执行分组试卷等各种任务，但仍然存在许多不完善之处，各种缺点仍然不可避免。还有许多未被发现的问题，各种问题必须在长期的运行和使用中逐步发现。通过不断发现问题和解决问题，可以使系统更加完善和实用。
参考文献
[1] Z. Liu, J. Liu, and Z. Liu, “Analysis, design, and implementation of impulse-injection-based online grid impedance identification with grid-tied converters,” IEEE Transactions on Power Electronics, vol. 35, no. 12, pp. 12959–12976, 2020.
[2] H. Sugisawa, T. Shinoda, Y. Shimizu, and T. Kumagai, “Cognition and implementation of disaster preparedness among Japanese dialysis facilities,” International Journal of Nephrology, vol. 2021, Article ID 6691350, 9 pages, 2021.
[3] A. Al-Aqbi, R. Al-Taie, and S. K. Ibrahim, “Design and implementation of online examination system based on msvs and SQL for university students in Iraq,” Webology, vol. 18, no. 1, pp. 416–430, 2021.
[4] S. Clivaz and T. Miyakawa, “The effects of culture on mathematics lessons: an international comparative study of a collaboratively designed lesson,” Educational Studies in Mathematics, vol. 105, no. 1, pp. 53–70, 2020.
[5] M. Mizuma, H. Yamamoto, H. Miyata et al., “Impact of a board certification system and implementation of clinical practice guidelines for pancreatic cancer on mortality of pancreaticoduodenectomy,” Surgery Today, vol. 50, no. 10, pp. 1297–1307, 2020.
[6] S. Makino and D. Lehmberg, “The past and future contributions of research on Japanese management,” Asian Business & Management, vol. 19, no. 1, pp. 1–7, 2020.
[7] Z. W. Lie, Q. L. Zheng, S. Zhou, and H. L. Rauf, “Virtual energy-saving environmental protection building design and implementation,” International Journal of System Assurance Engineering and Management, vol. 13, Supplement 1, pp. 263–272, 2022.
[8] Z. Tian, S. Tian, T. Wang, Z. Gong, and Z. Jiang, “Design and implementation of open source online evaluation system based on cloud platform,” Journal on Big Data, vol. 2, no. 3, pp. 117–123, 2020.
[9] J. Acosta, F. Amórtegui, A. Escobar, L. M. Leon, and S. Rivera, “Design and implementation of prototype for XLPE cable aging test,” Revista Internacional de Métodos Numéricos para Cálculo y Diseño en Ingeniería, vol. 36, no. 3, pp. 36–44, 2020.
[10] M. Cao, “Design and implementation of multidimensional interaction in online English course under the assistance of Omnimedia,” Scientific Programming, vol. 2021, Article ID 3713161, 10 pages, 2021.
[11] N. Choudhary, “Design and implementation of wildfire monitoring system,” International Journal for Modern Trends in Science and Technology, vol. 7, no. 5, pp. 139–143, 2021.
[12] K. Sharma and B. K. Chaurasia, “Trust based location finding mechanism in VANET using DST,” in Fifth International Conference on Communication Systems & Network Technologies, pp. 763–766, IEEE, Gwalior, India, 2015.
[13] A. Muneer and D. Zhan, “Design and implementation of automatic painting mobile robot,” IAES International Journal of Robotics and Automation (IJRA), vol. 10, no. 1, pp. 68–74, 2021.
[14] P. Elechi and C. O. Ahiakwo, “Design and implementation of an automated security gate system using global system for mobile communication network,” Journal of Network and Computer Applications, vol. 7, no. 1, pp. 1–10, 2021.
[15] S. Kaddoura, D. E. Popescu, and J. D. Hemanth, “A systematic review on machine learning models for online learning and examination systems,” PeerJ Computer Science, vol. 8, no. e986, 2022.
[16] L. Hu, “Design and implementation of a component-based intelligent clothing style CAD system,” Computer-Aided Design and Applications, vol. 18, no. S1, pp. 22–32, 2020.
[17] J. Li and W. Li, “On-line PID parameters optimization control for wind power generation system based on genetic algorithm,” IEEE Access, vol. 8, pp. 137094–137100, 2020.
[18] R. Dutta, A. Mantri, and G. Singh, “Evaluating system usability of mobile augmented reality application for teaching Karnaugh-maps,” Smart Learning Environments, vol. 9, no. 1, p. 6, 2022.
[19] B. Alhijawi, Y. Kilani, and A. Alsarhan, “Improving recommendation quality and performance of genetic-based recommender system,” International Journal of Advanced Intelligence Paradigms, vol. 15, no. 1, pp. 77–88, 2020.
[20] S. Ghareeb, A. J. Hussain, D. Al-Jumeily et al., “Evaluating student levelling based on machine learning model’s performance,” Discover Internet of Things, vol. 2, no. 1, p. 3, 2022.
[21] B. Xia, X. Zheng, L. Zhang, and L. Zhao, “UWB positioning system based on genetic algorithm,” Journal of Computer and Communications, vol. 9, no. 4, pp. 110–118, 2021.
[22] M. A. Elaziz, A. A. Ewees, and Z. Alameer, “Improving adaptive neuro-fuzzy inference system based on a modified salp swarm algorithm using genetic algorithm to forecast crude oil price,” Natural Resources Research, vol. 29, no. 4, pp. 2671–2686, 2020.
[23] A. Gholami-Rahimabadi, H. Razmi, and H. Doagou-Mojarrad, “Multiple-deme parallel genetic algorithm based on modular neural network for effective load shedding,” Soft Computing, vol. 25, no. 21, pp. 13779–13794, 2021.
[24] Z. Xu, M. M. Kamruzzaman, and J. Shi, “Method of generating face image based on text description of generating adversarial network,” Journal of Electronic Imaging, vol. 31, no. 5, article 051411, 2022.
[25] W. Xia and L. Shen, “Joint resource allocation at edge cloud based on ant colony optimization and genetic algorithm,” Wireless Personal Communications, vol. 117, no. 2, pp. 355–386, 2021.
[26] P. Rani, R. Kumar, A. Jain, and S. K. Chawla, “A hybrid approach for feature selection based on genetic algorithm and recursive feature elimination,” International Journal of Information System Modeling and Design, vol. 12, no. 2, pp. 17–38, 2021.
本文译自：Dengqing Z, Zhangwei Y. [Retracted] Design and Implementation of Online Japanese Examination System Based on Genetic Algorithm[J]. Wireless Communications and Mobile Computing, 2022, 2022(1): 3678607.