• 作者： 何正信; 周復華; 耿介信; 陳榮靜; 林志明; 許俊達
• 作者服務機構： 台灣工業繼續學院電子工程系
• 中文摘要： MBS為我們四年來研究廣義式黑板架構的成果。其知識源為一微黑板架構可提供多樣化的局部推理與知識表示方法，其黑板結構可任意對解題空間作水平與垂直方向的切割，其控制模組為一多黑板架構，包含一規範黑板子系統隨時提供解題指針，以及一環境黑板子系統負責找出領域黑板內的特質區和可執行的知識源以決定解題焦點。MBS容許系統設計師選定或細化上述三大部分以產生一適合特定領域的黑板架構。
• 英文摘要： This paper reports the result of our four-year research on the generalized blackboard architectureMBS. It employs generalized structures in its three components, i.e., knowledge source, blackboard struc-ture and control module. The knowledge source is designed as a micro-blackboard architecture to supportlocal representation and inference diversities. The blackboard structure is designed to support both homo-geneous and heterogeneous decomposition of a problem solving space. The control module is designed asa multi-blackboard architecture, including a constraint blackboard subsystem, to propose and update prob-lem solving guidance, and a context blackboard subsystem to identify distinct textures of the domainblackboard and to identify invocable domain knowledge sources to control the focus-of-attention. Most ofthe ingredients in these three generalized components can be selected or refined by the system designer sothat he can specialize the MBS into a specific blackboard architecture according to the domain requirementsto better fit his applications.
• 中文關鍵字： artifical ontelligence; blackboard architecture; knowledge-based systems
• 英文關鍵字： --