以Web為基礎的DNS專家系統之研製與實作

Abstract

隨著網路應用的多元化使得DNS的管理日趨複雜，因此要使DNS能夠有效地運作，必須要有豐富且專業的經驗才能達成，我們可以利用專家系統來輔助DNS管理。同時隨著網路技術的發達，專家系統已能建置在網路上，並以網頁的形式提供更友善及便利的服務。在這篇論文中，我們利用專家系統開發工具—DRAMA和其核心知識庫—NORM建置Web-based的專家系統，並且在網路上提供服務，以幫助DNS管理者解決問題。在建構知識庫的同時，我們提出了一套兩階段的知識擷取方法來以有效地從專家或個案中擷取出規則。在第一階段中，我們利用本體論的協助先建構出知識庫的架構，第二階段則是利用CommonKADS-like的模型分析法從個案擷取出規則，之後依照NORM的架構安排這些規則，並存放到知識庫中。在系統實作上，我們已經在網站上提供DNS問題診斷以及DNS系統檔案的自動化產生的服務。同時我們也利用問卷的方式由這些管理者提供意見來改善我們的系統，以期能夠使這套系統更加完善。

The applications of Internet are more and more complex nowadays and the tasks to manage DNS will become more and more difficult. Therefore, it will need expertise to do these tasks well. Using expert system approach to build an intelligent system is popular to support assistance for users. And as the popularity of Internet, the expert system technology has been applied to various applications in network services. Thus, we can build expert system on Internet to provide more convenient service. In this thesis, we build a web-based DNS expert system to assist DNS administrators in managing their DNS. When building the knowledge base, we propose a two-phase KA algorithm to fast elicit rules from experts or cases. In the first phase, we create the skeleton of knowledge base with the help of ontology; in the second phase, we use CommonKADS-like analysis modeling approach to elicit rules and then arrange these rules into the knowledge base. Finally, we develop DNS diagnosis and DNS configuration sub-systems on the web site to provide convenient service. And we adopt questionnaire approach to get the suggestions from the DNS administrators to evaluate the performance of our system.