臺灣博碩士論文加值系統

資料探勘(data mining)是一門運用廣泛的學問，在近幾年來由前人發展出許多不同的演算法，包括：Association rule、Classification、Clustering、OLAP、Decision tree、Neural network、Genetic algorithms[1]…等等。它的運用範圍幾乎涵蓋了各行各業，像是：生產製造、財務投資、信用卡交易、天文科技、服務業…等等。
本論文是將這門學科運用於半導生產製造，最主要是討論如何提升生產過程中的產品良率，其實在生產過程中影響良率的因素有很多，但是我們認為，其中一個很有可能的原因，就是由於某些機器的組合所導致，為了證明我們的思考方向，我們便設計出一套演算法來解決這個問題。
我們首先將在生產線上的產品生產過程中，每個步驟所經過的機器，以及一些量測的參數記錄下來。再透過我們所發展出來的演算法處理後，可以得到一些步驟的機器編號組合和一些經過運算後的參數，藉著它們可以用來輔助我們來判別哪些機器組合的生產良率特別的不好或特別好，以做為下次規劃生產排程時的參考。只要是生產良率特別不好的機器組合，在下次規劃生產排程時就應該儘量的避免將它們組合在一起，或者是儘量將好的組合排在一起。

Data mining is a wide-used knowledge. Many algorithms had been developed from the last years, such as Association rule, Classification, Clustering, OLAP, Decision tree, Neural network and Genetic algorithms [1] etc. The usage of data mining includes all trades and professions, such as the fabrication, financial investment, transaction of credit cards, astronomical technology and service trade etc.
We will discuss the usage of data mining in the fabrication of semiconductor. Although there are many reasons to affect productivity, we think that the associations of the machines will one of the reasons to affect the productivity. The main issue of our paper is how to use this knowledge to promote the productivity. In order to prove our idea, we design an algorithm to solve this problem.
First, we keep the record of product line, including every step of machine number and some other parameters and process the data with our algorithm. We could get the data that processing by our algorithm. With the data we can judge the association of machine that are special well or bad. Next time when scheduling, we could avoid the bad association, or take use of the well association.

第一章 簡介 1
第二章 相關背景 4
2.1 資料探勘(Data Mining) 4
2.1.1 資料探勘的流程 4
2.1.2 資料探勘的演算法 5
2.2半導體的製作流程 7
第三章 演算法說明 8
3.1 將半導體製程抽象化 8
3.2 資料模型 9
3.3 Cube的概念以及演算法的說明 10
第四章 實作方式 14
4.1以多維陣來解決問題 14
4.1.1 產生遮罩位元 15
4.1.2 將資料累計至cube中 16
4.2以一維陣列來解決問題 17
4.2.1 以一維陣列來當作快取記 18
4.2.2 將資料累計至cube中 19
4.3 效能分析 20
第五章 個案研究與結果分析 22
5.1 來源資料 22
5.2程式的運作說明 22
5.3程式執行結果及歸納分析 24
第六章 結論 26
參考資料 28
附錄一 來源資料表 30
附錄二 經過資料探勘後的結果資料表 42
附錄三 個案研究程式碼 45

[1] P. Adriaans and D. Zantinge, “Data Mining”, Addison Wesley Longman, 1996.
[2] 陳秋雲, “VLSI製造技術”, 全華科技圖書股份有限公司, 1999
[3] Cabena and Peter, “Discovering Data Mining: From Concept to Implementation”, Prentice Hill, 1998.
[4] H. B. Ajroud, “Automatic Architectural Abstraction”, North-Holland, pp.107~127, 1997.
[5] R. Agrawal, T. Imielinski, and A. Swami. Database Mining: A Performance Perspective. IEEE Transactions on Knowledge and Data Engineering, pages 914-925, December 1993.
[6] H. B. Ajroud and A. Jaoua, “Abstraction of Objects by Conceptual Clustering”, Journal of Information Sciences, pp.65~94, 1998.
[7] Berson and Alex, “Data Warehouse, Data Mining, and OLAP”, McGraw-Hill, 1997
[8] R. Srikant and R. Agrawal, “Mining Quantitative Association Rules in Large Relational Tables”, ACM SIGMOD, pp.1~12, 1996.
[9] D. Sceppa, “Programming ADO”, Microsoft Press, 2000.
[10] U. M. Fayyad, “Advances in Knowledge Discovery and Data Mining”, AAAI Press, 1996.
[11] W. R. Vaughn, “Hitchhiker’s Guide to Visual Basic and SQL Server Sixth Edition”, Microsoft Press, 2000.
[12] J. A Michael, Berry and Gordon Linoff, “Data Mining Technique: For Marketing, Sales, and Customer Support”, Wiley Computer Publishing, 1997.
[13] R. Agrawal, T. Imielinski, and A. Swami. Mining Association Rules between Sets of Items in Large Databases. Proceedings of ACM SIGMOD, pages 207-216, May1993.
[14] R. Agrawal, and R. Srikant. Fast Algorithms for Mining Association Rules in Lagre Databases. Proceedings of the 20th International Conference on Very Large Data Bases, pages 478-499, September 1994.
[15] M.-S. Chen, P. S. Yu. Database Mining for Path Traversal Patterns in a Web Environment. Proceedings of the 16th International Conference on Distributed Computing Systems, pages 385-392, May 27-30 1996.