| College students have strong autonomy in learning and life,so students are more susceptible to learning,living habits and emotional problems due to the surrounding environment and information.Therefore,student performance fluctuations and psychological barriers are more common in colleges and universities.With the advancement of university information construction,students’ data of all dimensions have been stored through the system.Therefore,using the massive data of students to extract the characteristics of students’ learning,life and psychology,and constructing clear student profile,it will improve the efficiency and pertinence of student management and realize early warning of student problems.Based on the research on the theory and technology of student profile,this paper designs and implements the university student profile system.According to the designed student data model,this paper obtains the learning,consumption,and forum data of 10022 students,and constructs the student profile labeling system according to the data model and system requirements.According to the system’s functional requirements and labeling system,this paper uses the fuzzy C-means algorithm optimized by simulated annealing algorithm to analyze the students’learning and consumption level,and uses text analysis algorithm to analyze students’ hobbies,topics of interest and sentiment index,and choose SVM algorithm pail.Students’ predictions were made.The Apriori algorithm was used to predict students’ subjects with risk.The C4.5 decision tree algorithm was used to analyze students’ life,learning and emotional abnormalities,and to alert students to the changes in student status.At the same time,the algorithm is optimized from the aspects of distance metric function of fuzzy C-means algorithm,initial cluster center determination,fuzzy index determination,cluster number determination,and perturbation function of simulated annealing algorithm.After systematic verification and testing,the clustering effect of the optimized fuzzy C-means algorithm is significantly improved,the accuracy rate of the warning for the subject is 83%,and the highest confidence level for the association rules for the student’s subject warning is 73%.The accuracy rate of the warning is 85%and the recall rate is 100%.The function and performance of the system meet the system requirements,and the label display is intuitive and the user experience is good. |
