新物件導向規則庫管理系統: 設計與實作

Abstract

在一個知識庫系統中，知識的更新是十分頻繁的。所以，一個具有高度可維護性的知識模型將可以有效的減少在知識庫管理上的負擔。在本論文中，我們試圖在可模組化、可再用性以及可分享性這三個主要層面上，提升一個知識庫的可維護性。因此，我們參考人類的學習及思考模式，提出一個新的基於物件導向的規則庫模型（NORM）來達成以上的三個特性。在這個模型中，三種主要的關連，包括知識間的參照，擴充及轉換，被用以描述分散的知識類別間的合作關係，進而提升知識的可維護性，尤其是在網際網路上的知識分享性。同時此知識模型亦整合了確信因子模式以支援具有不確定性的推論行為。根據NORM知識模型，我們藉由XML定義出一套相對的知識表示語言，ORML，並且發展出一套對應的物件導向規則庫管理系統。除了基本的知識管理及推論功能外，這個系統亦支援分散式的架構。最後，一個適性化電腦輔助教學系統，將被用來說明如何利用我們的規則庫管理系統開發一個完整的知識庫應用系統。

In a knowledge base system, the knowledge base is updated frequently. Therefore, a knowledge model with high maintainability can reduce the effort of managing a knowledge base. In this thesis, the maintainability includes three criteria: modularity, reusability and sharability. According to learning and thinking mode of human, New Object-oriented Rule Base Model (NORM) is proposed to achieve above features. In this model, three main relationships, reference, extension and transference, are used to describe the cooperation of distinct classes of knowledge concept. Hence, NORM improves the maintainability of knowledge base, knowledge sharing over Internet especially. Besides, certainty factor model is adopted in NORM to support the uncertainty reasoning. Based on NORM, a knowledge representation language, Object-oriented Rule Markup Language (ORML), is defined using XML and a corresponding system, Object-oriented Rule Base Management System (ORB-MS) is developed. Excluding basic facilities of managing and inferring, the system supports the distributed architecture. Finally, an E-learning system, individualized computer-assisted learning system, is developed to describe how to develop a complete knowledge base application using ORB-MS.