以知識為基礎之軟體元件管理模式探討

Abstract

一如個人電腦(PC)、筆記型電腦(NB)等硬體產業走向分工製造模式之際，近年來，台灣的軟體產業也試圖建置起一個完善的分工體系，以收專業化的實效；故現階段大型軟體專案的開發漸以多人、異地的團隊，在共同的軟體基礎架構上來進行。軟體業者分工的想法和硬體業者如出一轍，但在發展成果尚有一段的差距；透過Web Service技術，現階段之研究已能初步建立起以軟體元件為基礎之軟體產業的水平分工整合模式，然而在軟體元件的管理上（如元件的儲存、搜尋及檢索等)，仍有很大的改善空間。\r\n採行軟體元件分工模式來組裝、開發資訊系統，意味著系統整合廠商需能方便地取得符合其功能需求之軟體元件。基於此，若能建置一元件儲存庫，並提供完善的儲存、搜尋及檢索機制以提供系統整合廠商找尋其所需元件，應能提升軟體元件分工的成效。而一個良好的元件儲存庫應具備一系列完善的綱要(Schema)，用來進行元件分類，並詳述儲存於元件庫內之元件的特性。\r\n本研究之目的即在於發展出一個能支援軟體產業分工的軟體元件管理模式；以元件式軟體供應鏈水平整合為基礎，提出一軟體元件之分類綱要，及知識導向之元件儲存庫架構來儲存、搜尋、並檢索適用的軟體元件，同時，也期望透過元件知識推論專家系統的發展，來提升元件檢索之效益。另一方面，元件選擇流程中的元件評估階段也是本研究所要探討的主題。

In recent years, the software industry in Taiwan attempts to build a complete system of division of labor to obtain the effectiveness of specialization as the computer hardware industry did in implementing the mechanism of supply chain management successfully. The idea of software component supply chain is similar to the concept of supply chain model of hardware industry. Although some models of horizontal integration of software industry have already been proposed, management mechanisms of software components are seldom considered in these models (ex: storage, searching, and retrieval).\r\nTo effectively adopt the software component supply chain to achieve the horizontal division of labor, the system integration (SI) companies should be able to access software components, which conform to their functional requirements, conveniently. Therefore, if we can develop a component repository and a mechanism for storing, searching and retrieving the components in the repository.\r\nThe purpose of this research is to develop a software component management model, which can support the division of labor in software component supply chain, based on the component-based software industry horizontal integration. Firstly, we will propose a descriptive schema to describe and to classify the software components. Secondly, we will use this schema to design a knowledge-based component repository, which is used for storing, searching and retrieving suitable software components. Also, we develop an expert system for inferring component knowledge in order to improve the performance of component retrieving. Finally, the component evaluation in the process of component selection will be important subjects to be addressed in this research.