Friendship Network Mining And High-level Behavior Analysis Based On Campus Big Data

With the in-depth development of education informatization and internet education,the construction of digital campuses in the universities is rising day by day.While intelligent devices are convenient for students to live on campus,they also record a large amount of student behavior data.The popularization and application of educational big data mining and artificial intelligence have laid a solid theoretical and application foundation for us to mine valuable information from massive data.Due to the vigorous development of wireless network WiFi and smart phone APPs,as well as the popularity of the Campus Digital Management System(CDMS)and wireless Radio Frequency Identification(RFID)system,we are able to easily obtain and analyze the trajectory and behavior data generated by students 'colorful campus life.Mining the valuable information can help education managers provide more scientific and effective management of data support guidelines.At the same time,to effectively use student behavior data and analyze behavioral characteristics have certain practical significances for individualized quality education,intelligent management,and teaching quality and work efficiency development.This thesis is based on the student's smart campus card,wireless WiFi,and campus digital management system to collect student behavior data,including student shopping,dining,consumption and book borrowing,library access control,dormitory,bathing,Internet access,and student performance information.Although these data are messy and noisy,they contain unintuitive but very valuable information.First,this thesis will use these data to build friendship networks,analyze their composition and structure,investigate the relationship between students' academic performance and friendship networks through social network analysis.Thus,by finding the positive factors of the friendship network,can help us better understand the student's school life and personalized development.Second,calculate and analyze student behavior from three high-level behavioral characteristics,i.e.,orderliness,diligence and memory,and obtain the relevant characteristic values or behavior regularity of students.And use correlation analysis to study the relationship between behavioral characteristics and academic performance,revealing the impact of daily behavioral characteristics of students on academic performance,which can help educators discover abnormal students,guide students to improve bad behavior habits,and then improve students' academic performance.The results of the study show that the orderliness,diligence,and memory characteristics have significant relationships with academic performance,which are positively correlated with academic performance.The distribution of orderliness has a Gaussian,while the diligence and memory show a power-law distribution;the higher the orderliness of behavior or diligence is,the better the academic performance.There is a significant heavy-tail effect on the distribution of time interval of behavior.These studies can provide guidance for the construction of student behavior models,help educators understand the behavioral characteristics,and ultimately benefit most students.