Design And Research Of Online Examination System Of Test Questions Recommendation Based On Neural Graph Model

The examination is a particularly important link for the whole learning process of students.However,in the process of manual question setting,paper printing,student examination,manual marking and score statistics,traditional paper examination will incur a large amount of manpower and time expenditure.Moreover,the test questions are fixed or random and lack of individual pertinence.To solve the problems of traditional paper-based examination,this paper designs and develops an online examination system.At present,online examination systems have been widely used,but these systems often ignore students' personalized learning needs.The existing examination system in the market,mainly for the purpose of making profits,pays more attention to the aesthetic design of the interface and the expansion of functions;Many colleges and universities have their own test systems,which are designed to measure students' recent learning,so they pay more attention to the core online test function,which is relatively simple.In this era of huge data and information overload,Recommendation system as a personalized solution,has been widely used in various fields.Personalized learning methods can better stimulate students' interest in learning,more targeted to help students learn and improve their performance.In order to realize students' personalized learning needs and meet the needs of the school and the platform,this paper has realized the functions of automatic paper formation,marking,and the function of simulating test paper,practicing knowledge points and redoing wrong questions.At the same time,the recommendation algorithm is introduced into the developed examination system.This paper designs two ways to select questions in the test exercise function module,which are the traditional random selection and the recommendation algorithm.For the function of topic recommendation,this paper studies a kind of efficient recommendation algorithm,which can be used for the recommendation of test questions.According to the history records of students' wrong questions,students can be recommended some questions that are easy to be wrong and difficult to master.The experiment also shows that the recommendation algorithm studied in this paper has improved the recommendation effect compared with other similar algorithms.The main work of this paper is as follows:(1)According to the needs of the school and the platform,an online examination system has been developed.And for students' personalized learning needs,this paper studied and understood some typical applications in the field of recommendation system,and explored the principles of classical traditional recommendation algorithms such as content-based recommendation,model-based recommendation and collaborative filtering.(2)For test question recommendation exercise,the recommendation algorithm is studied.A recommendation algorithm model NGCF-Att based on deep learning and graph structure is proposed,which is based on the neural graph collaborative filtering algorithm(NGCF).Firstly,the inner product is replaced by the multi-layer neural network in the interaction layer of NGCF.Secondly,attention mechanism is introduced in the message construction of the propagation layer.(3)Tensorflow was used to implement the recommendation algorithm,and experiments were carried out on public data sets such as Amazon-book and Gowalla that could be mapped into the database format of this system to verify the effect of the recommendation algorithm.