新產品研發（New Product Development）是一連串與各式各樣包括過往經驗與重要文件等知識相關的過程。在現今以文件為主的知識管理系統（Knowledge Management System）與產品資料管理（Product Data Management）系統中檢索文件，大多以關鍵字或使用者的權限為主要依據。一方面必須能在使用者執行現在工作時提供與此階段相關的知識，另一方面一旦使用者缺乏該領域的知識或甚至是剛入門的新手，將無法透過錯誤的關鍵字來檢索適合的知識來協助使用者。 本研究致力於發展一個整合式的架構，能夠整合數個支援新產品研發整體生命週期知識管理的資訊系統。為了能夠管理過往經驗以及相關文件等知識，我們發展了與工作、階段、分類、角色以及元件相關的知識模型，來表現出新產品研發流程中的相關知識。並且為了記錄個別經驗，建立了以延伸標記語言（eXtensible Markup Language, XML）為基礎的案例結構，該結構不僅記錄了每個案例的基本資訊，也提供了使用該案例找尋參考資料的方法與專家等推薦的其他參考資料。以及希望能夠檢索適宜的推薦案例，我們也整合了案例式推理（case-based reasoning）與使用知識階層的合作式查詢問答（cooperative query answering）等機制。 最後，由於新產品研發整體生命週期中的前兩個階段：產品概念設計與產品細部設計階段，需要參考到大量並分散的知識。我們將提出的架構方法以應用到此兩個階段為例，來驗證本研究提出的各種方法。 New product development (NPD) is series of processes which involve multidisciplinary knowledge including past experiences and other critical documents. Retrieving the docu-ments in current document-based knowledge management system (KMS) and Product Data Management (PDM) system based on several keywords and/or permission of the users. On the one hand it is necessary to provide knowledge actively depends on the stage of NPD process where the users perform currently, on the other hand once the users’ lack of sufficient domain knowledge or even a novice, they are not able to retrieve the adaptive knowledge via incorrect keywords. This paper aims to develop an integrated architecture which coordinates several infor-mation systems supporting NPD total lifecycle with knowledge management. In order to manage the knowledge of past experiences and related documents, we develop knowledge model which involves the tasks, phases, classifications, roles, and components of product development describing the representation of knowledge in NPD processes. For recording the individual experience, we establish XML-based case structure which not only records the basic information of the experience but also the utilization of the case including the recom-mendation of experts and experienced users. On the purpose of retrieving the adaptive rec-ommendation, we also integrate the case-based approach and cooperative query answering techniques with knowledge hierarchy. For validation the proposed approach, we apply to two examples which are in the early phase of NPD total lifecycle involving product concept and detail design where is required for referring a great deal of scattered knowledge.
