結合資料倉儲與資料探勘的技術分析中小學數位落差

Abstract

對於一個商業或研究機關團體在從事主題研究後，累積了大量的研究資料時，要如何有效的管理及善用這麼龐大的重要資源，成為每一位研究工作者所要面對的課題。本論文主要在提出如何應用「資料倉儲」(Data Warehousing) 及線上分析處理(On-Line Analytic Processing, OLAP)的技術，完成「應用資料倉儲技術之問卷分析」的架構與設計，並利用「資料探勘」(Data Mining)技術來對「中小學數位落差資料」進行分析與歸納的研究。 研究過程可概分為三個階段(1)資料前處理階段:將蒐集到的資料進行過濾、整合、轉換等資料前處理程序後，成為可適用於資料倉儲的資料格式。(2)資料倉儲建置階段:將資料前處理程序後的資料，建立成具有多維度資料模型結構的資料方塊體（Data Cube）後，存入「中小學數位落差資料倉儲」。(3)線上分析及資料探勘階段:在多維度資料倉儲建置完成後，便可進行「線上分析」及「資料探勘」的處理，產生有意義的資訊或特徵。而且，使用「資料探勘」的技術，結合受訪者的相關背景特徵資訊來進行群集(Clustering)分析，利用分群後的群集個體差異，來建立可代表各群集的「數位學習落差形成因素」決策樹(Decision Tree)，再從決策樹歸納出有效的規則，可供學生、教師、學校作進一步的調查或研究之用，也可作為推行中小學資訊教育計畫決策的依據或相關研究工作的使用。

To conduct researches of some specific topics, we should firstly collect related resources. Therefore, how to manage and use such enormous and important resources becomes an issue to deal with. In this thesis, we will bring up the ideas of how to apply Data Warehousing and On-Line Analytic Processing techniques to carry out the framework of this research, and then make use of Data Mining techniques to analyze the data resources of The Digital Divide of K-12. There are three phases in the research process including the preprocessing phase, the data warehousing phase and the OLAP and Data Mining phase. In the preprocessing phase, the raw data will be filtered, transformed and integrated into the suitable format for the data warehouse. In the data warehousing phase, a multidimensional data cube will be built based on the preprocessed data from prior phase, and then will be stored in the Data Warehouse of The Digital Divide of K-12. In the OLAP and Data Mining phase, after the multidimensional data warehouse has been built, the OLAP and Data Mining procedure can be performed and some meaningful results may be generated. Also, some Data Mining techniques can be applied to perform cluster analyses on the background of the interviewees. Finally, the differences of the clusters can be used to build the Decision Tree that represents the factors which form The Digital Divide of K-12. The effective classification rules extracted from the Decision Tree will help students, teachers or schools for further investigation. These results may be useful for making policy decision for the development of information education in K-12 or other related researches.