A Recommendation Algorithm For College Entrance Examination Based On Graph Neural Network

College Entrance Examination Major Recommendation(CEEMR)is an auxiliary mechanism to help students quickly and accurately understand their preference for Major,so as to help students choose the majors they apply for through the auxiliary mechanism.In China,college entrance examination is related to a student's future career development,and has always been a hot topic of public concern.Therefore,college entrance examination specialty recommendation has a broad prospect.Due to different national conditions,foreign scholars rarely recommended to participate in the professional field,therefore the development of professional recommendations,is to promote domestic scholars,mainly domestic existing professional recommendation algorithm based on collaborative filtering?fuzzy FCM and other traditional recommendation algorithm,which based on collaborative filtering recommendation algorithm,the best it is through the students interested in professional grading data to match similar students,according to the choice of interest students have similar professional recommendation,but the score can't involve all professional students,therefore more professional without the corresponding score data can cause data sparse,at the same time if the student in the number of samples is less,Failure to match students with similar interests will result in cold start of data.To solve the above problems,this paper proposes a recommendation algorithm for college entrance examination majors based on graph neural network combined with the popular graph convolutional network.The main work is as follows:1.The first data set based on class was established,and the collected data set was established as graph structure.Overcoming the traditional questionnaire which takes students as individuals,the innovation of choosing class as unit in this paper is that it is convenient to integrate students' class background and students' social relations,etc.,and alleviate the problems of sparse data and cold start of data by adding friends' scores on majors.2.The graph neural network model was proposed for the first time to solve the problem of major recommendation in college entrance examination.Due to the influence of the interest is mutual,each student's characteristics are not isolated,it should consider social interaction between students and friends,and just can put the figure structure around the node aggregation to the current node,through the integration of students evaluation and the surrounding friends information,the extraction of more complete student interest and professional characteristics.3.The effectiveness of the graph neural network model is verified.Compared with the professional collaborative filtering personalized recommendation algorithm,the MAE evaluation index is reduced by 10%,and the RMSE is reduced by 21%.The model is verified by an ablation test.The rationality of the model framework.