- 學位論文
以生產規則知識架構設計採購決策支援系統之研究
Decision support design of purchase system with production knowledge structure
摘要
企業在進行採購時，採購人員扮演重要角色。本研究以晤談採購人員進行知識擷取，規劃得出21 項採購考量因素。進而在企業既有ERP 中，以採購考量因素運用層級分析法(AHP)設計生產規則，以建構採購決策支援系統。層級分析法中的權重可由採購人員依經驗進行設定，以符合AHP 的一致性要求。 實務上，生產規則採用SQL Command來實現規則式表示法(Rule Based)。在決策支援系統運算後，供應商積分將以「K-Means」和晤談之採購人員所提出「指定合格數」和「總績分數值」進行分群。將分群結果給予評語，並繪製簡易統計圖提供給決策者，可協助決策者明確、快速之決策報告畫面。 知識庫分為「規則庫」、「模式庫」及「決策報告資料庫」。規則庫可由決策者以實際需求進行組合，AHP權重設定及運算都在規則庫進行。模式庫是記錄21項採購因素轉換成對應SQL Command，實際上，模式庫是用來運算Raw Data。決策報告資料庫除記錄決策報告資料，還記載決策過程中所產生之數據，過程中數據可供決策者檢視更細微的決策軌跡。 本論文實作一個採購決策支援系統，實驗資料引用某企業西元2012年之ERP資料，並請該企業兩位採購人員進行AHP權重設定，再加上Dickson(1966)和Wilson(1993)所提出評選供應商指標，共四位專家之評選重要性數據進行實驗。實驗數據再運用Precision和Recall進行分析，從中發現決策建議之方式，影響Precision和Recall的範圍及分佈圖。實驗結果發現，採購人員A的Precision和Recall分佈圖和Dickson的分佈圖較為相近，且問卷調查結果也顯示，採購人員A比較重視的評選因素也與Dickson提出的因素較為相符。
並列摘要
Purchasing specialists (PSs) play an important role in the process of purchasing in the enterprise. This study acquires knowledge by consulting two PSs to get 21 significant purchasing factors. Based on these purchasing factors, analytical hierarchy process (AHP) is used in some existing ERP system to design a set of production rules for decision support systems. The weightings in the AHP are adjusted by PSs according to their heuristics so as to meet the consistent requirement of AHP. In practice, the production rules use SQL commands to perform rule-based representation. After the calculation of the supplier's scores, the decision support system finds clustering results by three methods: K-Means, "PS-proposed Threshold" and "total score", then outputs suggestions to the decision makers. Finally, a simple statistic chart will be drawn to provide decision makers a clear report. This study builds a knowledge base which consists of a rule base, a model base and a decision report base. The rule base is made up by decision makers according to practical requirements. The AHP weightings are set and stored in the rule base. The model base works on raw data to transform 21 purchasing factors into corresponding SQL commands. The decision report base records not only decision report data but also the intermediate data, which provides decision makers more elaborated decision-making processes. This decision support system was put into practice. The experiment data came from an enterprise's ERP data set in 2012. Two PSs of this enterprise are consulted to build two sets of AHP weightings. The "supplier evaluation criteria", proposed by Dickson (1966) and Wilson (1993), are also applied to make another two sets of AHP weightings. We compared the outcomes of these four weighting sets. The experiment results are justified by precision and recall, and the suggested decision-making methods can be determined. The experiment results show that the precision-recall distribution of PS A is similar to the Dickson's, and the results of the questionnaire show that the factors that PS A considered as important are similar to the criteria proposed by Dickson.
參考文獻
延伸閱讀
- 郭勃顯、杜瑩美、張盛鴻、李榮貴（2008）。新產品專案組合決策支援系統之研究。明新學報，34(1)，107-126。https://doi.org/10.29688/MHJ.200802.0008
- 王俊中、許秉瑜、蕭文龍、謝淑慈（2010）。全球供應鏈產銷規劃之決策支援架構：決策模型與演算法。臺大管理論叢，20(2)，135-171。https://doi.org/10.6226/NTURM2010.20.2.135
- 吳慧蘭（2021）。應用設計思考理論建構資訊戰情中心支援生產單位快速決策之研究-以A公司為例〔碩士論文，中原大學〕。華藝線上圖書館。https://doi.org/10.6840/cycu202100155
- Tsai, H. Y. (2008). 以群體決策理論解析採購評選作業之研究 [doctoral dissertation, National Taipei University of Technology]. Airiti Library. https://doi.org/10.6841/NTUT.2008.00546
- 張嘉昇（2008）。Design Course Enrollment and Management for Decision Support System using Large Item Sequence〔碩士論文，中原大學〕。華藝線上圖書館。https://doi.org/10.6840/CYCU.2008.00383