The Application Of Educational Intelligence In M-learning Scenario

Data mining has emerged to be one of the most vivacious areas in information technology in the last decade. Compared with the booming fact in academia, data mining applications in the real world has not been as active, vivacious and charming as that of academic research. This runs in the opposite direction of Data mining's original intention and its nature. Unfortunately we are not at the stage of holding as many effective algorithms as we need in the real world applications. What we need is its application in practice and promting its research on application in various field.Data on various aspects of education field is rapidly increasing. These vast amounts of educational information data is collected, stored in a wide variety of data container, so a growing number of educational researchers transfer their focus on how to change these massive data into useful information quickly. Based on understanding Data mining history and related applications in the field of education and inspired by the business intelligence, this paper presents "Educational Intelligence" concept (educational knowledge-driven data mining applications) which is an opportunity to look forward to the flourish of the applied research in this area, and eventually integrated with the education and research methodology.The main work done by this thesis includes three parts:Firstly, a comprehensive analysis of the field of education at home and abroad (including mobile learning domains) status of applied research in data mining, introduced the data mining concept and methodology , described several methods of data mining, cited common data mining tools and discussed the the main evaluation factors of data mining tools. Base on these, this paper put forward the concept of educational intelligence inspired by business intelligence, then discussed the distinction and relationship between educational intelligence and educational statistics.Secondly, in m-learning scenario,mined and analyzed the data which used two cases to show the mobile learning domain knowledge-driven data mining application. In'The Application of Decision Tree Technology in Gender Differences Research of Mobile Learning'case,based on the large scale original experimental data samples form mobile learning practice, the article uses C5.0 algorithm to respectively generate boy and girl groups'courseware-used-satisfaction decision tree, and then compare extracted dissatisfied rules to study the gender differences. In'The Application of Cluster analysis methods to Evaluate the Usage of Small-Chunk Learning Resources'case, it reports our efforts in designing and developing"small-chunk"learning resources which were based on knowledge points and experimentally applied in Mobile-Learning. After learners finish one stage of Mobile-Learning by using these learning materials, we collect survey data and apply Data-Ming technology to execute a cluster analysis on large scale data which produced a more valid result. By defining characteristics of various style learners and changing rule among them, research analyzes the feasibility and detailed design strategy of small-chunk self-contained learning sources in Mobile-Learning.Finally, in the mobile learning system,based on CRISP-DM model, designed and implemented a time-series prediction Web service to predict the download demands on website resources. This paper put emphasis on presenting the design conception and key technologies on implementation of CRISP-DM Model for time-series prediction task. The test result proved that the time-series prediction Web service renders fairly high prediction accuracy, fast deployment and easy to use. It is of considerable referential importance in solving similar problems.