博士生毕业因素的关联规则挖掘:群像特征与类型差异  

An Exploration of Association Rules for Doctoral Graduation Factors:Group Image Features and Type Differences

作　　者：王婷 耿有权[2] 黄亚继 WANG Ting;GENG Youquan;HUANG Yaji(Graduate School,Southeast University,Nanjing 210096,China;Research Center for Learning Science,Southeast University,Nanjing 210096,China)

机构地区：[1]东南大学研究生院,南京210096 [2]东南大学学习科学研究中心,南京210096

出　　处：《重庆高教研究》2024年第5期98-112,共15页Chongqing Higher Education Research

基　　金：江苏省学位与研究生教育教学改革项目“大数据驱动下的研究生学位质量动态监测与风险预警研究”(JGKT23_C004)。

摘　　要：博士生教育以培养具有学术创新能力的高层次人才为目标,肩负着知识生产和促进经济社会发展的重任。在大数据背景下,利用博士生培养过程的教育数据,深度挖掘博士生毕业时间与培养各环节的关联规则,发现博士生延期毕业的主要原因,探析不同时间毕业的博士生群体特征,可为高校加强博士生培养管理、提高博士生毕业率提供科学指导。基于某高校研究生管理大数据系统,获取学生入学信息、导师基础信息、综合素质表现、校园行为情况、科研学习情况5类数据,处理整合形成数据集,并进行特征提取和标签体系构建,采用FP-Growth算法挖掘所有标签与博士生毕业因素的关联规则,最终刻画出按期毕业、延期毕业和超期毕业3类博士生的群体画像,并对比分析3类群像呈现的差异性特征。研究发现:博士生入学方式、课程成绩、开题情况、科研成果数量和完成时间、在读期间是否获得奖学金和荣誉表彰、毕业论文送审时间与博士生毕业时间存在强关联性;博士生入学年龄、入学前院校是不是“双一流”高校、导师年龄、导师任博导年限、导师指导学生人数、校园行为习惯与博士生毕业时间关联性不强;按期毕业和超期毕业博士生分别对应一种典型群体画像,延期毕业博士生包含3种典型画像;3类博士生群体画像在学术志趣、学术基础和学术能力3个方面呈现明显差异。为提升博士生培养质量,提高博士生毕业率,需打破刚性限制与隐性歧视,确保优秀生源选拔的有效性;强化过程监督与分流退出,确保培养过程管理的科学性;重视课程供给与基础培养,保障科研实践衔接的流畅性;呵护博士生学术志趣和身心健康,激发其科研探索和创新的积极性。Doctoral education aims at cultivating high-level talents with academic innovation ability,shouldering the important task of knowledge production and promoting economic and social development.In the context of big data,the educational data of doctoral training process was used to explore the association rules between doctoral graduation time and training links,discover the main reasons for delayed graduation of doctoral students,and analyze the group characteristics of doctoral graduates at different times,providing objective scientific guidance for strengthening the management of doctoral training and improving the rate of doctoral graduation in universities.Based on the acquisition of five types of data from the management big data system of graduate students in a university,namely student enrollment information,basic information of supervisors,comprehensive quality performance,campus behavior data,and research and learning situation,data processing was integrated to form a data set,feature extraction and label system construction were carried out,and FP-Growth algorithm was used to explore the association rules of all labels and doctoral graduation factors.Finally,the group portraits of three kinds of doctoral students who graduated on time,graduated after the deadline and failed to graduate after the deadline were depicted,and the different characteristics of the three kinds of group portraits were compared and analyzed.The results show that there is a strong correlation between the admission method,course score,proposal situation,the number and completion time of scientific research results,whether they won scholarships and honors during study,and the graduation time of graduation thesis.The age of doctoral students,whether the university is“double first-class”before entering the university,the age of the supervisor,the years of the supervisor as a doctoral supervisor,the number of students under the guidance of the supervisor,the campus behavior and the graduation time of doctoral students are not stron

关 键 词：博士生教育 博士生群体画像 按期毕业 延期毕业 超期毕业 

分 类 号：G643[文化科学—高等教育学]