本研究運用資料探勘技術，使用R語言關聯規則、文字探勘套件分析中華電信公司電子採購資料，並將結果提供做為MRO類型的採購作業精進的方向，以利組織達成敏捷性與企業成本優化目標。MRO採購資料品項繁雜、非結構化資料不易分析、解讀，現在可透過快速發展的大數據、資料探勘演算法、視覺化分析工具，更進一步的掌握採購資料的樣貌，精進採購管理作業。
MRO採購作業自需求提出、請購、審核、產品與供應商的資訊搜尋、比價、議價、下單、付款等流程，近年來因逐漸受到企業重視，已陸續於各階段導入電子化採購，結合資訊流、金流、物流的E化流程，已大幅減化作業流程、提高採購效率，但MRO採購過程中的人力物力成本、庫存成本、物料價格等顯性成本，以及由於供貨不及時、不準確而導致營運績效不彰，甚至服務中斷等的隱性成本，在企業採購總成本中仍被視為成本節降、需精進管理的重點。在企業微利時代，做好MRO採購管理工作已受到企業的重視。
本研究先利用關聯規則演算法分析出中華電信公司電子採購資料的關聯規則與頻繁項目集，並由結果得知中華電信公司財物、勞務、工程三大類採購類別，僅工程類採購案較能掌握採購品項為「營繕工程」、「機械工程」，財物及勞務類採購案內容多出現「其他財物」、「其他勞務」的不明確狀況。接著再利用文字探勘詞頻分析、語意模型分析採購短文資料，找出小額採購之財物、工程、勞務共三大類別採購案，各自的採購關鍵字，進而分析出較頻繁採購的品項，提供管理階層由不同面向獲取採購精進標的資訊。建議可從所獲得之前八十大關鍵字詞集中，進行逐年的 TF-IDF分析，追蹤其趨勢是逐年上升或逐年下降，可合理預測採購品項及供需面趨勢。

This study uses data mining technology, association rule mining in R language, and text mining to analyze e-procurement data of Chunghwa Telecom Company. The results are used to improve MRO procurement operations and enable organizations to fulfill the purposes of agility and cost optimization. The data of MRO procurement items is complex and unconstructed, making it difficult to analyze and interpret. Now, with the rapid development of big data, data mining, and visual analysis tools, we can have a better knowledge of the procurement data, and improve the management of procurement operations.
MRO procurement operations include the process of requesting, purchasing, reviewing, information searching of products and suppliers, price comparing, price negotiating, ordering, and paying. In recent years, this process has received increasing attention by companies and has been gradually introduced into e-procurement. In combination with the e-process of information flow, cash flow, and logistics, this process has reduced the operating process substantially and has improved the efficiency of procurement. However, explicit costs, such as labor and material costs, inventory costs, and material prices in the MRO procurement process, as well as hidden costs, such as poor performance and service interruption due to insufficient or inaccurate supply, are highlighted. They are regarded as the targets of cost reduction and further management, in the total cost of procurement. In the era of micro-profits, management of MRO procurement has been valued by companies.
In this study, association rule mining algorithms are used to analyze association rules and frequent itemsets of Chunghwa Telecom Company’s e-procurement data. From the results, it is learned that among Chunghwa Telecom Company’s three major procurement categories, property, labor service, and construction work, only cases of construction procurement, with procurement items of “construction and maintenance work” and “mechanical engineering,” can be better controlled. Cases of property and labor service procurement often have ambiguous situations, with procurement items of “other property” and “other labor service.” Next, term frequency analysis in text mining and semantic model are used to analyze procurement data, to find four major procurement methods, including limited bidding, open bidding, inter-entity supply contract, and small procurement, and their respective procurement keywords. Those can further be used to analyze the items that are purchased more frequently and provide managers with information on procurement targets from different aspects. From the top eighty keywords obtained, TF-IDF analysis is performed year by year. By tracking whether the annual trend is increasing or decreasing, procurement items and trends of supply and demand sides can be reasonably predicted.

中文文獻
1.丁再興,2002,MRO電子市集採購關鍵成功因素之研究,臺北科技大學商業自動化與管理研究所碩士論文,民91年。
2.范錚強,2019.02 updated,SCM:Supply Chain Management電子化供應鏈管。
3.謝邦昌,2017,SQL Server資料採礦與商業智慧：適用SQL Server 2016/2014(第二版)。
4.沈彥廷,2016,臺北醫學大學生物統計研究中心eNews第14期。
5.廖述賢,朱佩慧,2019,以文字探勘與書目分析法探討資料探勘技術的發展與應用,德霖學報第32期。
6.陳宗權,陳俊育,2016,運用文本探勘技術探索未來科技,科技部105年度自行研究計畫成果報告。

網路部分
1.CRISP-DM 1.0用戶指南。
2.MBA智庫百科,MRO採購。
3.Microsoft Power BI,2020。
4.陳嘉惠,鄭順林,跨產業資料探勘標準作業程序(CRISP-DM)在心臟病資料分析之應用。
5.郭耀仁,2018,R語言與資料科學應用。

英文文獻
1.Agrawal,R.and Srikant, R. (1994) Fast Algorithms for Mining Association Rules in Large Databases. Proceedings of the 20th International Conference on Very Large Data Bases, Santiago de Chile, 12-15 September 1994,487-499.
2.David M.Blei,2012,Probabilistic Topic Models,communications of the acm,April 2012,vol.55,no.4
3.Fayyad,U.G.,G.Piatetsky-Shapiro and P.Smyth,1996,The KDD process for extracting useful knowledge from volumes of data,Cummunication of the ACM,39(11).
4.Gaikwad,S.V.,Chaugule, A.,&Patil,P.,2014,Text mining methods and techniques. International Journal of Computer Applications,85(17),42-45.
5.Hadley Wickham,Garrett Grolemund,2017,R for Data Science,Oreilly.
6.Jiawei Han, Micheline Kamber,Jian Pei,2011,Data Mining:Concepts and Techniques,3/e.
7.Spasić,I.,Ananiadou,S.,McNaught,J.,&Kumar,A.,2005.Text mining and ontologies inbiomedicine: Making sense of raw text. Briefings in Bioinformatics,6(3),239-251.
8.Struhl,S.,2015,Practical Text Analytics:Interpreting Text and Unstructured Data for Business Intelligence:Kogan Page Publishers.
9.Tan,P.-N.,Steinbach,M.,&Kumar,V.,2006,Introduction to Data Mining: Addison-Wesley Longman Publishing Co.,Inc.
10.Turban, E.,Sharda,R.,&Delen,D.,2013,Decision Support and Business Intelligence Systems:Pearson Education,Limited.