臺灣博碩士論文加值系統

資料萃取是從儲存大量資料的資料庫中篩選過濾、探索出具有新的意義之關連性、特徵與趨勢的技術。近來隨著網際網路的普遍化，資料的收集也越來越方便，更顯得資料萃取（Data Mining）的重要性。因為資料的數目會隨著電子化時代的來臨日漸膨脹，想要從資料庫中萃取出有用的資訊，變得更為困難。在這種情形下，資料萃取為一最佳選擇。但是由於資料萃取過程中需要不同領域的專家參與，但專家的判斷常伴隨著主觀的意見，容易產生誤判的情形發生，而這些誤判會造成企業在決策過程中，造成重大的損失。
本篇論文因此提出三種提供模糊歸屬函數的機制，來取代資料萃取過程中專家的判斷，避免標準的變更。方法一是利用類神經網路中的二維SOM網路將多維的資料產生聚類，方法二是利用類神經網路中的一維SOM網路將單維的資料產生聚類，方法三是結合一維SOM與K-means法，利用兩階段法將單維的SOM法所求出的聚類數與群組重心放入K-means法內，重新計算群組重心與範圍，所發展的自動聚類法。藉由群組重心建構模糊三角函數，再配合以模糊集合為觀念的模糊演算法，產生最終的關聯性法則。最後，本研究採用一現實生活中的資料庫，來求取此三種聚類法在資料萃取過程所產生不同關聯性法則的數目與信賴值。

It is a modern trend for an enterprise to use computers in every business process. The result is that huge amount of enterprise data is collected by computers. Data have to be analyzed effectively so that useful enterprise knowledge can be retrieved and utilized. But past technologies cannot serve for this purpose. Data mining is a new technology aiming at transforming the raw data into valuable information. In the process, different domain experts are needed to provide different information. For most of fuzzy data mining researches, the fuzzy membership function needs to be provided by the domain experts.
In this thesis, three approaches are provided to assist in deriving the fuzzy membership functions. We use the two-dimensional and single-dimensional SOM (self-organizing map) neural networks, and a combination of the SOM network and the K-means method to determine the appropriate number of groups for data attributes. When the group centers that are the appropriate number of groups for data attributes are decided, these centers are used to construct the triangle fuzzy membership functions. Next, the fuzzy association rule algorithm is used to retrieve the fuzzy customer behavior knowledge. In the process, the support and confidence values are used to filter out the noise values and unimportant attributes. Experiments are performed to evaluate all the approaches. Raw data from a library are examined and the fuzzy customer behavior knowledge is retrieved.

摘要 i
ABSTRACTii
誌謝iii
目次iv
表目錄vi
圖目錄vii
第一章 緒論1
1.1 研究背景與動機1
1.2 研究目的3
1.3 研究限制3
1.4 研究步驟與流程4
1.5 各章節的概述5
第二章 文獻探討6
2.1 資料萃取基本概念6
2.1.1 類神經網路在資料萃取中的應用9
2.1.2 統計學在資料萃取中的應用14
2.1.3 以Itemset為基礎的資料萃取應用15
2.2 模糊理論18
2.3 聚類的方法21
2.3.1 K-means 聚類法22
2.3.2 自組織映射圖網路聚類法23
2.3.3 兩階段聚類法25
2.3.4 其他聚類法26
第三章 研究方法27
3.1 資料的收集29
3.2 自動化聚類法29
3.2.1 SOM聚類演算法30
3.2.2 SOM+K-means 聚類演算法33
3.3 模糊歸屬函數的制訂35
3.4 模糊資料萃取法35
3.4.1 符號的定義36
3.4.2 模糊資料萃取演算法36
第四章 實證結果40
4.1 多維SOM結合模糊資料萃取演算法的實證結果41
4.2 單維SOM結合模糊資料萃取演算法的實證結果46
4.3 單維SOM+K-means結合模糊資料萃取演算法的實證結果50
4.4 比較結果54
第五章 結論與建議58
5.1 結論58
5.2 研究貢獻59
5.3 後續研究建議59
參考文獻61
附錄
A 資料萃取軟體畫面65
B Apriori演算法運算過程68
C 原始與分析資料資料庫內容71

[1]王嘉隆，資料倉儲應用中的兩層資料庫間的邏輯設計方法，國立中央大學資訊管理研究所， pp.1-15，民國85年。
[2]吳育儒，決策數中移除不相關值問題的研究，淡江大學資訊工程系碩士論文，民國87年。
[3]張博欽，模糊概念圖中資訊量應用於資料探索之研究，中原大學資訊工程系碩士論文，民國86年。
[4]鄭志宏，異質性資訊源環境中之資料採掘：以屬性為基礎之方法，國立台灣科技大學電子工程技術研究所碩士論文，民國87年。
[5]H. Maurice and S. Arun, “ Set-oriented Data Mining in Relational Databases”, Data &Knowledge Engineering, Vol. 17, pp.245-262, 1995.
[6]U. Fayyad and P. Stolrz, “Data Mining and KDD: Promise and Challenges”, Future Generation Computer Systems, Vol. 13, pp.99-115, 1997.
[7]M. S. Chen, J. Han and P. S. Yu, “Data Mining: An Overview From A Database Perspective”, IEEE Transactions on Knowledge and Data Engineering, Vol.8 No.6, pp.866-883, December 1996.
[8]W. Anita, M. Ernestina and H. Mike, “A Generalization Model for Implementing a Data Mining System”, Seventh National Conference on Fuzzy Theory and Its Applications, pp77-81, 1999.
[9]M. D. June, C. S. Jack, C. H. Gregory and L. S. Jr. Richard, ”Mining multi-dimensional data for decision support”, Future Generation Computer Systems, Vol. 15, pp.443-441, 1999.
[10]S. S. Anand, A. R. Patrick, J. G. Hughes and D. A. Bell, “A Data Mining methodology for cross-sales”, Knowledge-Based Systems, Vol.10, pp.449-461, 1998.
[11]Michael J. A. Berry, Data Mining Techniques for Marketing, Sales and Customer Support, (Wiley Computer Publishing, 1997).
[12]C. S. Kuo, T. P. Hong and S. C. Chi, ”A study of fuzzy data mining algorithms for quantitative values”, Graduate School of Management Science I-Shou University Thesis, 1999.
[13]C. Y. Lee, T. P. Hong and Q. R. Zeng, “Automatic Acquisition of Fuzzy Knowledge”, Graduate School of Electric Machinery Chung-Hua University Thesis, Taiwan, 1995.
[14]S. H. Rubin, “A Fuzzy Approach Towards Inferential Data Mining”, Computers and. Engine, Vol.35 Nos.1-2, pp.267-270, 1998.
[15]T. P. Hong and C. Y. Lee, “Effect of Merging Order on Performance of Fuzzy Induction”, Intelligent Data Analysis, Vol. 3, pp.139-151, 1999.
[16]H. T. Hsieh and Y. R. Syau, “Fuzzy Numbers in Financial Credit Rating —An Approach to Bank Credit Policy”, Seventh National Conference on Fuzzy Theory and Its Applications, pp.49-52, 1999.
[17]T. P. Hong, C. S. Kuo and S. C. Chi, “A Data Mining Algorithm for Transaction Data with Quantitative Values”, Seventh National Conference on Fuzzy Theory and Its Applications, pp.874-878, 1999.
[18]盧瑞山，類神經網路於環境資訊之鑑識、推估及預測之研究，國立台灣大學含竟工程學研究所博士論文，pp.124-133，民國88年.
[19]H. Lu, S. Rudy and H. Liu, “Effective Data Mining Using Neural Networks”, IEEE Transactions on Knowledge and Data Engineering, Vol.8 No.6, pp.957-961, December 1996.
[20]S. George, “Seeking Signals in Data Bases using Neural Networks”, Fifth International Symposium on Signal Processing and its Applications, pp.197-170, 1999.
[21]W. C. Mark and W. S. Jude, ”Using neural networks for data mining”, Future Generation Computer Systems, Vol.13, pp.211-229, 1997.
[22]D. Alahakoon, S. K. Halgamuge and B. Srinivasan, “Data Mining with Self Generating Neuro-Fuzzy Classifiers”, IEEE International Fuzzy Systems Conference Proceedings, pp. II-1096-1101, August 22-25, 1999.
[23]C. C. J Fu and E. S. Lee, “Neuro-Fuzzy in the Valuation of Stock Price Based on Fundamental Analysis”, Seventh National Conference on Fuzzy Theory and Its Applications, pp57-60, 1999.
[24]楊敏生，類別資料之模糊聚類分析，中原大學數學研究所碩士論文，pp.20-25， 民國88年。
[25]尋勇智，應用層次及群分析法評估電力系統運轉模擬，逢甲大學電機工程研究所碩士論文， pp.55-65，民國85年。
[26]胡承民，自組織映射圖神經網路與多變量集群分析法在市場區隔之應用﹍以3C賣場為例，義守大學管理科學研究所碩士論文，pp21-36，民國87年。
[27]W. C. Hou, “Extraction and Applications of Statistical Relationships in Relational Databases”, IEEE Transaction on Knowledge and Data Engineering, Vol.8 No.6, pp.939-945, December 1996.
[28]W. Beat, “Probabilistic Knowledge Bases”, IEEE Transactions on Knowledge and Data Engineering, Vol.7 No5., pp.691-698 , October 1995.
[29]E. Age, J. Robert, L. Thomas and S. L. Clark, “Data Mining and Neural Networks for Knowledge Discovery”, Nuclear Instruments and Methods in Physics Research, A 389, pp.251-254, 1997.
[30]V. Juha and A. Esa, “Clustering of the Self-Organizing Map”, IEEE Transactions on Neural Networks, Vol.11 No.3, May 2000.
[31]R. Agrawal, I. Tomasz and S. Arun, “Database Mining: A Performance Perspective”, IEEE Transactions on Knowledge and Data Engineering, Vol.5 No.6, December 1993.
[32]R. Agrawal, I. Tomasz and S. Arun, “Mining Association Rules between Sets of Items in Large Databases” ACM SIGMOD Conference Washington DC, USA, May 1993.
[33]J. M. Peña, J. A. Lozano and P. Larrañaga, “An Empirical comparison of four initialization methods for the K-Means algorithm” Pattern Recognition Letters 20, pp.1027-1040, 1999.
[34]R. Agrawal, J. Gehrke, D. Gunopulos and P.Raghavan, “Automatic Subspace Clustering of High Dimensional Data for Data Mining Applications”, Proc. of the ACM SIGMOD Int’l conference on Management of Data, Seattle, Washington, June 1998.
[35]L. N. de Castro and F. J. Von Zuben, “An Improving Pruning Technique with Restart for the Kohonen Self-Organizing Feature Map”, IEEE, pp.1916-1919, 1999.
[36]D. S. Sim, “Self-Organizing Neural Networks For Unsupervised Color Image Recognition”, IEEE, pp.338-342, 1991.
[37]蔣以仁，資料發掘之模糊分類，國立台灣大學資訊工程學研究所博士論文，台北，pp.143-146，民國86年。
[38]http://www.almaden.ibm.com/cs/quest/
[39]http://www.sinter.com.tw/SPSS/spss_product.htm
[40]http://www.sas.com/technologies/data_mining/index.html
[41]H. T. Lefteri and E.U. Robert, Fuzzy and Neural Approaches in Engineering, (John Wiley & Sons Inc., 1997).
[42]Krzysztof Cios, Witold Pedrycz and Roman Swiniarski, Data Miming Methods for Knowledge Discovery, (Kluwer Academic publishers, 1998).
[43]J. M. Peňa, J. A. Lozano and P. Larraňaga, “An empirical comparison of four initialization methods foe the K-Means algorithm”, Pattern Recognition Letters, Vol. 20,pp.1027-1040, 1999.
[44]R. J. Kuo, L. M. Ho and C. M. Fu, “Integration of Self-Organizing Feature Map and K-Means Algorithm for Market Segmentation,” Computers and Operations Research. (Accepted).
[45]G. Punj and D.W. Steward, ”Cluster analysis in marketing research: review and suggestions for applications”, Journal of Marketing Research, Vol.20, pp.134-148, 1983.
[46]R. Agrawal, J.C. Shafer: "Parallel Mining of Association Rules", IEEE Transactions on Knowledge and Data Engineering, Vol. 8, No. 6, December 1996.
[47]R. Srikant, R. Agrawal: "Mining Quantitative Association Rules in Large Relational Tables", Proc. of the ACM-SIGMOD 1996 Conference on Management of Data, Montreal, Canada, June 1996.
[48]R. Srikant, R. Agrawal: "Mining Generalized Association Rules", Proc. of the 21st Int''''l Conference on Very Large Databases, Zurich, Switzerland, Sep. 1995.