網際網路原是為了快速交換科技資訊而建立的平台，但教育學者將網際網路運用在課程的教學上，提供所有人能夠不分時間、不分地點，即時需求的學習方式，不僅改變了教師授課的方式，更改變學員學習的方式。近年來雲端的興起，資訊安全教育開始利用雲端的主要技術虛擬化（Virtualization）保存與還原當時的環境，不僅易管理，更節省大量的建置成本，因此我們利用虛擬化與代理人(Agent)技術打造CSEP雲端安全實驗平台，並放置多元化的互動式案例供學員操作。 但要將一般的展示案例轉換成具有互動效果之互動式案例不僅需要花費許多的心力完整了解案例本身程式碼的架構，更需要花費許多時間在適當的地方放置互動點，隨著未來展示案例複雜度的增加，所耗費的時間可預期的將會直線上升，因此本研究依據過往製做互動式案例的經驗，針對PHP網頁程式碼，分析目前CSEP雲端安全實驗平台與額外數個樣本，並歸納出一系列之規則機制未來供教材設計者遵循。為了要驗證本研究之方法有效性，我們額外挑選測試樣本並比對過往經驗與本研究提出之機制兩者挑選出的互動點，最後利用Precision、Recall與F-measure衡量本研究方法的正確度，實驗結果顯示這些歸納出的規則具有於各層面皆有40%以上的Precision與，65%以上的Recall表示本研究提出之規則雖需耗費較多的搜尋成本，但找出的結果足以涵蓋絕大部分的可行解，最後在F-measure的評估也具有53.33%以上的正確性，而Web語意分析的F-measure介在49.8~78.79%和Document Clustering的F-measure介在30%~80%，故印證了以這些規則檢測是可有效用於互動點。As internet is a platform for exchanging information, education researchers applied it to provide real-time demand learning. This application changes not only teaching but also learning. The rise of cloud computing in recent years pushes this change. The virtualization technology in cloud computing enable us to manage easily and save lots of cost for constructing experiment environment. We have built a cloud security experiment platform (CSEP) to provide instructive experiments with interactivity for trainees. However, transforming a normal security case into an interactive case would require much time to understand the structure of vulnerable source code and put interactive point into appropriate place. To reduce this burden, we propose rules which are concluded from the analysis of five CSEP cases and fifteen PHP projects to recommend the place to set interactive points.To prove the effectiveness of our proposal, we choose other PHP projects and compare interactive points which are selected by exporters and our method respectively, then use precision, recall and F-measure to measure the performance of our method. The result shows the precision rate, recall rate and F-measure are more than 40%, 65% and 53% respectively. Considering the F-measure in Web semantic analysis is between 49.8% and 78.78% and in document clustering is between 30% and 80%, our method provides a promising approach to recommend interactive point in interactive E-learning cases.
