案例式推理在工程變更知識推薦應用

Abstract

在製造業的生產過程中，負責生產的工廠裡，往往會有不斷地改善生產流程、提高生產率或降低成本的BKM(Best Known Method) 或CIP (Continue Improvement Program) 專案發生。然而，這樣的BKM/CIP在跨廠區或全球性的公司中，經驗的萃取與分享不是那麼容易取得，導致相同的好或壞的驗證實驗計畫不斷地在不同廠區被執行。相對也造成資源的浪費，增加了生產成本，而且跨區域的製程知識無法及時且有效的被引用，改善計畫的執行期間可能更加的壟長。如何能將有用的資訊從繁複的日常工程變更中整理和正確的歸類，以便能找出最合適或相近的處理案例來推薦給工程師，以加速工程變更的處理，是重要的研究議題。 案例式推理主要是運用先前的案例，作為解決日後問題的參考依據。案例式推理應用的範圍相當廣泛，包括信用風險預測、顧客流失、客服系統、IT服務管理…等等的應用。若能結合資料探勘的技術，將有效率地從大量資料中整理出有用的資訊。本研究建構一個雛型系統針對所提方法做一比較驗證。藉由文件自動分類方法與案例式推理技術，萃取出隱藏在大量工程變更中的寶貴經驗，達成經驗的推薦及知識的分享，進而降低工程改善的成本，並強化公司整體競爭力。

In manufacturing, there are many BKM (Best Known Method) or CIP (Continue Improvement Program) of engineering change tasks executed to improve product’s quality, yield or cost reduction. However, the BKM/CIP cases couldn’t be referenced easily for a world-wide company. The repeating CIP tasks might be executed in different fabs waste resource and increase manufacturing costs due to the valuable BKM/CIP cases not well managed and shared between cross fabs. It also might lead a long cycle time in the following CIP tasks. It is an important issue for enterprises to extract useful information from complicated engineering change process in everyday and correctly categorize cases and find out the most suitable cases for speeding up improvement tasks.

The Case-Based Reasoning (CBR) is the process of solving new problems based on the solutions of similar past problems. CBR is widely used such as credit risk evaluation, customer loss situations analysis, customer service, IT helpdesk service, etc. This work integrates CBR with Text Mining technology to extract useful information from text-based documents. Use Text Mining technique to categorize cases, and then we use CBR to look for the most similar cases in selected category to recommend to engineers in new engineering change cases solving. A prototype system is implemented to demonstrate the effectiveness of our approach. This work will extract valuable information and support engineers to easily get similar cases in engineering change process flow and solve problem efficiently and quickly and it will also strengthen the enterprise competitiveness.