- 學位論文
以本體論為基之多代理人系統以支援需求鏈管理-以IC設計股務為例
Ontology-based multi-agent system for Supporting Demand Chain Management: the Case of IC Design Services Firms in Taiwan
摘要
隨著快速變化的市場需求及資訊科技的發展，企業逐漸重視需求鏈管理(Demand Chain Management)所帶來的效益並視之為提升競爭力的新議題。這使得企業可以獲得最新的市場需求資訊以便及時開發需求產品，滿足市場需求。因此，需求鏈上的成員必須溝通和合作以達到結合需求資訊、縮短前置時間及減少成本。 本研究是以台灣IC設計服務公司為背景，探討以本體論為基之多代理人系統如何支援需求鏈管理並促使跨公司團隊之間的溝通與取得及時需求資訊。多代理人系統可以讓來自不同公司的代理人在特定的領域裡以本體論理的架構與規則，直接進行有效的協調合作。需求鏈參與成員之間緊密的連結關係呈現一個相輔相佐的關係，彼此之間的需求知識交流也為了達到瞭解與回應需求的目的以創造更大的利益價值。此研究中所提出的需求鏈管理本體論呈現企業回應顧客需求過程中所需的多層級知識架構及其關聯性，開發團隊可以透過代理人系統整合獲取所求的資訊，並且應用在研發與生產階段，以便可以及早回應。 研究中的案例主要是探討如何以本體論為基之多代理人建構的系統雛型來促進彼此之間的知識分享及應用將需求知識，以便合作團隊易於找尋所需要的資料，澄清與滿足顧客需求、降低成本及前置時間；也根據案例需求設定雛型中的代理人的特質如溝通規則、行為和決策等，希望能幫助需求鏈裡的上下游公司的溝通與合作。 In the fast changing market environment, Demand Chain Management (DCM) has been considered as a powerful strategy to increase enterprise competency. This practice enables enterprises to get the latest market demand information for further product and process development to meet market needs in the early stage of development This research explores the issue of how ontology-based Multi-Agent System (MAS) help to facilitate the communication of demand information among IC design service companies. MAS is expected to enable relevant agents participating team members from different companies to coordinate and communicate based on the roles and rules specified in the ontology. In this research, agents are designed to represent department or person in-charge for quick response without waiting for instruction. The specific roles and rules for upstream and downstream partners in demand chain enable them to have systematic interaction to evaluate and balance the supply-demand conditions in order to create customer value. The DCM ontology in this research needs to represent multi-level knowledge structure and relationship among partners in demand chain for the partners to acquire needed knowledge through MAS in the development and production stage. Specifically, the case study in this research demonstrates the demand knowledge sharing among the agents in systematic format such as communication protocol, behavior and decision patterns in order to help the key performance in DCM such as: meet customer demand, reduced development cost and lead time.