Research On Educational Applications Of Key Learning Analysis Technologies Under The Background Of Big Dat

Big data is applied in the education to realize data-driven education process description,diagnosis,prediction and decision feedback,so as to analyze the educational problems,improve educational methods and improve educational quality.So that the value of massive educational data can be brought into full play and effectively fed back is a research hotspot for today’s educational practitioners and researchers.In recent years,the application research of big data driven education reform has begun to focus on two issues.How to design more efficient and effective methods from the laws and decisions of education in the context of big data?How to shift from the top-level design and theoretical analysis of big data and learning analysis to the innovation and application of key technologies and methods,rather than the local analysis and special measurement of specific cases with existing tools?About the above two problems,this dissertation refers to the relevant achievements of educational application researches of learning analytics at home and abroad,these achievements have demonstrated the importance and urgency of mining data value with key learning analytics,but not have not formed a unified framework applied to massive educational data,nor have they realized the innovation and design of key technologies in the framework,nor have they formed a implementation system for educational practice.Therefore,we construct the theoretical model of learning analytics,and set up four key components of the learning analytics framework,namely,descriptive learning analytics,diagnostic learning analytics,predictive learning analytics and instruction based learning analytics,they are used to solve the key problems of "what happened→why did it happen→what may happen→what needs to be done" in educational data.Firstly,we focus on the descriptive data analysis needs of online teaching process,about the problems faced in educational research in this field,we take a large data set of online learning behavior as the research object,the technical design and application of "learning behavior description analysis in the same period" and"learning behavior description analysis in the different periods" are carried out.It is used to solve the key research problem of "what happened" in the online teaching processes.Secondly,based on the large data set of online learning behaviors,supported by the relationships and conclusions obtained from the descriptive analysis,we provide methods and conclusions to support for the prediction analysis and instruction analysis,put forward the research issues about semantic feature selection,fusion and topology,and study three-dimensional application for the diagnostic analysis,they are:(1)Selection:sparse learning strategy and diagnostic selection of key semantic features;(2)Fusion:interactive response diagnostic recognition of context semantic feature fusion;(3)Topology:diagnostic learning strategy and topology coverage of semantic rules.It is used to solve the key research problem "why does it happen" in the online teaching processes.Thirdly,based on the big data set of online learning behaviors,and supported by the results and conclusions obtained from descriptive analysis and diagnostic analysis,considering the intelligent technologies in educational application research,we put forward the research issues,from point to line,from line to plane,and study three-dimensional application for the predictive analysis needs,they are:(1)Point:learning behavior prediction based on improved random forest algorithm;(2)Line:prediction of potential learning behaviors based on association rules;(3)Plane:random field design and cooperative prediction strategy of learning behaviors.It is used to solve the key research problem of "what may happen" in the online teaching processes.Finally,based on the big data set of online learning behaviors,supported by the results and conclusions of descriptive analysis,diagnostic analysis and predictive analysis,considering groupness and individuality,we put forward the research issues for the instruction based analysis of online learning behaviors,they are:(1)Group Decision:the decision of regression analysis of multi category learning behaviors;(2)Individual Decision:local sensitive decision based on interest transfer of learning interaction activities.It is used to solve the key research problem of "what needs to be done" in the online teaching processes.The key learning analytics in the context of big data will be used for education on the premise that valuable conclusions or basis can be obtained from the data,that depends on appropriate and accurate technical means,namely effective learning analytics.Education optimization driven by big data is the follow-up research direction.The application research of key learning analytics is to transfer data to values,that is the key issue to realize the comprehensiveness and practicality of future education and intelligent education.