高校贴吧中学生行为数据的文本聚类应用及研究

发布时间：2018-10-21 13:28

【摘要】：自打互联网出现以来,人们的工作生活方式都发生了翻天覆地的变化。在“互联网+”模式的新环境下,互联网不断地渗透到各个领域中。互联网应用的不断发展,积累了海量的网络行为数据,产生了“大数据”。随着“数字化校园”到“智慧校园”相关概念的不断提出,国内几乎所有高校都逐步推出了自己的校园网站、微信公众号和智能办公平台等各种教育教学服务系统,同时高校也成为了“大数据”应用的阵地。网络应用在大学生群体中同样占有首要地位,校园外各应用平台上也记录了海量的学生行为数据。学生行为对学生的学业成绩、学习状态都有影响。有效的引导学生行为是促进学生全面发展的必要条件。因此,高校教育管理者经常会对学生行为进行分析,以便及时发现问题、及时纠正。传统的学生行为分析主要以收集学生行为数据,利用数理统计等方法进行行为分析。在“互联网+教育”的新形势下,学生行为数据的产生已经不局限于校园,早就通过网络遍布世界各地。学生在校园内外各平台上应用都会留下记录,学生行为就已经可以很直接的表现在校园内外各平台中了。因此,教育管理者可以直接通过各平台收集学生行为数据进行分析,然后做出管理决策。由于各应用平台中记录的学生行为数据呈现多样化的特点,更多的是以诸如图片、文字、音视频等非结构化的数据形式存在。对教育管理者的分析工作带来了挑战和机遇,现代教育、教学管理发展的一个新趋势就是要综合考量多方面的评价指标来衡量教育、教学管理工作水平。对于校外应用平台上的学生行为数据,本课题选择网络社区中比较有代表性的平台,全球最大的中文社区----百度贴吧。百度贴吧运营多年,积累了相当大的数据量。在高校贴吧类目中同学们往往就同一个话题各抒己见,这些信息对学校管理决策很有参考意见。贴吧上同学们都是用文本进行阐述,产生的数据都是半结构化的。本课题针对百度贴吧数据的真实、客观、全面等特点,使用Weka开源工具,采用大数据中的数据挖掘技术在文本聚类应用中的K-Means和EM方法成功地对百度贴吧网站上收集到的学生行为数据进行文本聚类挖掘,以此来辅助教育管理者的学生行为分析工作。本文的愿景是希望以本课题的研究为铺垫,能够为教育管理者在学生行为分析工作中提供一种新的思路。
[Abstract]:Since the emergence of the Internet, people's way of work and life have undergone earth-shaking changes. In the new environment of "Internet" mode, the Internet continuously permeates into various fields. The continuous development of Internet application, accumulated massive network behavior data, produced "big data". With the development of the concept of "digital campus" to "intelligent campus", almost all colleges and universities in China have gradually launched their own campus website, WeChat public number and intelligent office platform and other educational and teaching service systems. At the same time, colleges and universities have also become "big data" application of the position. The network application also occupies the primary position in the university student group, the massive student behavior data is also recorded on each application platform outside the campus. Student behavior has an effect on students' academic achievement and learning state. Effective guidance of student behavior is a necessary condition to promote the overall development of students. Therefore, college education administrators often analyze students' behavior in order to find problems and correct them in time. Traditional student behavior analysis is mainly based on collecting student behavior data and making use of mathematical statistics to conduct behavior analysis. Under the new situation of "Internet education", the generation of student behavior data is no longer confined to the campus, and has been spread all over the world through the Internet for a long time. Students in and out of the campus platform will leave a record of the application, student behavior can be very direct performance in the campus and other platforms. Therefore, education managers can collect student behavior data directly through each platform to analyze, and then make management decisions. Because of the diversity of student behavior data recorded in various application platforms, it exists in the form of unstructured data such as picture, text, audio and video. The analysis of education managers has brought challenges and opportunities. A new trend in the development of modern education and teaching management is to comprehensively consider a variety of evaluation indicators to measure the level of education and teaching management. For the students' behavior data on the off-campus application platform, this topic chooses the representative platform in the network community, Baidu Tieba, the largest Chinese community in the world. Baidu Tieba operation for many years, accumulated a considerable amount of data. The students often express their opinions on the same topic in the post bar category of colleges and universities. This information is very useful for school management and decision making. Posted on the students are using text to elaborate, the data generated are semi-structured. This topic aims at Baidu Tieba's data true, objective, comprehensive and so on characteristic, uses the Weka open source tool, The K-Means and EM methods of big data's data mining technology in text clustering application are successfully used to mine the student behavior data collected on Baidu Tieba website, so as to assist the student behavior analysis of educational administrators. The vision of this paper is to provide a new way of thinking for education managers in the work of student behavior analysis.
【学位授予单位】：西华师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G647

【参考文献】