Study On Students Friend Relationship Based On Campus Multi-source Data

Under the background of promoting informationization and digital construction in colleges and universities,a large amount of student data is generated and accumulated.Fully exploring the potential correlation and value of these data can help school teachers to carry out student work and realize personalized service more scientifically.Good interpersonal communication and good relationship among students on campus are the basis for the development of college students' mental health,the cultivation of social communication skills and the healthy and happy learning on campus.It is also an important indicator to measure students' mental health.Therefore,based on the multi-source data in the school card system,this paper studies the relationship between students and campus friends.The Gaussian similarity function and PageRank improved algorithm are used to complete the friendship status,class cohesion construction and student campus activity evaluation.The research analysis aims to understand the situation of students under the line and to find out the group of students with low school activity and suspected isolation,and provide data support for the school's related management,so as to guide students to make healthier friends.This paper mainly carries out research work from the following four aspects:1)Based on the current research and research on relevant theories and techniques in data mining and campus big data research,this paper proposes the research topics,as well as the research methods and main core algorithms to be adopted;2)Based on the data of offline swiping card of college students' one-card,this paper puts forward the design of the model of "meeting" students,and speculates the situation of making friends on campus so as to better understand the status of students' life in school.In order to accurately calculate the correlation between students,the Gaussian similarity calculation method is introduced,Gaussian similarity analysis is carried out on the time and place information generated by students using campus card consumption and entering the library,and the correlation value between students is calculated,and then each student's offline "friends" or the research object is inferred.Students who have similar habits of living and working,while understanding the life status of students,can also provide friend recommendations for students in need;3)The traditional page importance ranking algorithm,PageRank algorithm,is introduced into the analysis of campus data environment.According to the research content of this paper,the traditional PageRank algorithm is improved accordingly.An activity analysis algorithm based on campus student relationship matrix is proposed.Combining the strength of the student relationship value with the probability matrix used in the algorithm calculation,the weight distribution of the probability matrix is optimized,and the method of distributing the node weights in the traditional algorithm is optimized,which is beneficial to get closer to the real result.In addition,the relationship matrix used in the research of this topic is the campus-card data accumulated by the student group at the same time,which overcomes the shortcoming of the PageRank algorithm “unfriendly to the newly added page” in the webpage ranking research.4)The experimental verification and result analysis of the feasibility of the algorithm are carried out,and the results are physically verified with the help of the relevant departments of the university involved in the data source.In the survey of sample individuals,the probability that the first friend of the student calculated by the ActivityRank model is the real friend of the research subject can reach 91%.In the analysis of the class cohesiveness situation,the correlation value between the class students is considered to be greater.The greater the number of people,the better the cohesiveness.Taking the relationship network of the two classes as a comparison,the class of students who are rated as“Top Ten Classes”has a higher degree of intensive relationship and student relevance than the student network of the ordinary class.In the analysis of the lower activity groups in the campus,the results of the study are also in line with the real situation of the school.At the same time,there will be "familiar strangers" in the offline "friends" judgment of students,that is,the two students have similar living habits but do not know each other.Therefore,the analysis of the consumption data of the data card can also provide a platform for the students in need to be recommended by their friends.By recommending students with similar living habits to them,the student can expand his social circle and improve the campus life and learning status of the group with low activity.