結合案例推理與專家找尋之問題診斷支援機制

Abstract

如今先進製程控制與資訊科技深深影響著產品發展的各個面向。隨著產品越加複雜與全球化生產趨勢，其應用之支援資訊系統分工亦越趨精細、複雜。而企業有鑑於此，針對其投入之人力、時間與金錢成本亦越趨龐大。然而對於資訊人員流動快速、專業知識與異常處理經驗無法有效儲存、利用等問題仍舊是一項重要待解的議題。 在此，本研究嘗試以案例推理架構為主體結合文字探勘技術與專家找尋方法，來有效儲存、利用半結構化案例資訊，藉此解決案例庫儲存、利用困難等問題，並以更有效率的方式取得貼近問題本質之案例資料供處理人員以為新產生問題處理之依據。最後，並輔以基於社群網絡分析為基礎之專家找尋機制，建議經驗專家予處理人員諮詢，進而有效補足案例資訊不足之處。 最後，本研究將針對所提之相關方法學實作雛型系統進行驗證，藉此期許能對企業內部資訊系統異常處理能力能有所提升，並強化其整體競爭力。

Nowadays advanced manufacturing and information technologies have impacted on every aspect of product development significantly. With the increasing of product complexity and globalization manufacturing, enterprise manufacturing information change more and more complex. Therefore, sharing manufacturing knowledge understanding such as personal experience and exception handling is one important issue to be solved. This paper introduces a research integrated text mining technique and top-k support documents expert search model based on Case-based reasoning (CBR) by using knowledge of the semantic structure of documents. CBR can use known experiences to solve new problems, we store the past problems as cases in a case base and a new case is classified by determining the most similar case from the case base. And the use of degree centrality of the expert candidate network can find the expert with the most influence through the experts score rank. Our approach of the CBR based expert recommends combined the reliability and influence between the experts can help users to get the expert support and resolve the potential problems of CBR such as lack of feedback and lack of a sufficiently rich case library. With these methodologies, we establish the Problem Support and Expert Recommend System (PSERS), which applied in the case base of fault diagnosis expert system of manufacturing information. From the experimental results, these techniques are shown to be very effective in the modeling and extraction of the domain knowledge in the case base.