關鍵查核事項(Key Audit Matter, KAM)的揭露，係希望讓投資人清楚理解公司本身存在的風險事項，以減少投資人與公司之間的資訊不對稱。我國自2016年開始分為兩階段施行審計準則公報第57號「財務報表查核報告」至今，關鍵查核事項的內容是否對於投資人確實具有投資方面的參考價值，為本研究想探討的核心。
本研究針對2016年到2018年我國的上市與上櫃(含興櫃)公司，採用文字探勘(Text Mining, TM)方式，分析每篇關鍵查核事項文字內容，量化關鍵查核事項內容傳達的營運風險(Business Risk)程度，以基因演算法(Genetic Algorithm, GA)優化支援向量機(Support Vector Machine, SVM)偵測公司營運風險。最後，以十折交叉驗證法(10-fold cross validation)來評估模型是否有過度擬合(overfitting)的情形。
結果發現關鍵查核事項確實具有反映公司營運狀況的資訊內涵，可作為投資人於制定投資決策時的參考。因此，本研究支持推動揭露關鍵查核事項，能達到維護投資人權益與健全資本市場。

The disclosure of key audit matters (KAM) is that investors can clearly understand the risks of the company to reduce the information asymmetry between investors and the company. Taiwan has implemented the Standards on Auditing No. 57 since 2016. The purpose of this study is whether the key audit matters can help investors to make the decisions on the investment.
This study is aimed at Taiwan's companies, including listed companies, over-the-counter (OTC) companies, and emerging stock companies, from 2016 to 2018. It uses the method of text mining (TM) to analyze the text content of each key audit matters to quantify the degree of business risk conveyed by the key audit matters, and uses the Genetic Algorithm (GA) optimizes Support Vector Machine (SVM) to detect the company's business risks. Finally, use 10-fold cross validation to assess whether the model has overfitting.
The results of this study indicate that the key audit matters have information about the company's operating conditions and can be used as a reference for investors in making investment decisions. Therefore, the results of this study support the promotion of the disclosure of key audit matters, which can protect investors' interests and improve the capital market.

圖目錄 i
表目錄 ii
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研究流程 4
第二章 文獻探討 5
2.1舊式查核報告的資訊價值與新式查核報告的資訊價值 6
2.2我國關鍵查核事項研究之現況 7
2.3營運風險的偵測與關鍵查核事項 9
2.4文字探勘技術 11
2.5分類模型 13
第三章 研究方法 16
3.1利用關鍵查核事項偵測營運風險之流程 16
3.1.1 建立營運風險特徵語料庫 17
3.1.2 營運風險之偵測 20
3.2研究資料 22
3.2.1 資料來源與研究期間 22
3.2.2 按關鍵查核事項類別分類 23
3.2.3 負評比率 28
3.2.4 合併關鍵查核事項資料集與企業信用風險指標(TCRI)資料集 28
3.2.5 關鍵查核事項類別變數 30
第四章 研究結果與分析 36
4.1偵測營運風險模型 36
4.1.1 資料型態 36
4.1.2 特徵標準化 37
4.1.3 預測結果 37
4.1.4 評估偵測營運風險模型 39
4.2其他機器學習方法 43
4.2.1 其他機器學習方法的資料型態 43
4.2.2 其他機器學習方法的特徵標準化 44
4.2.3 其他機器學習方法的預測結果 44
4.2.4 評估其他機器學習方法的模型 47
第五章 結論與建議 51
5.1研究結論 51
5.2研究限制 51
5.3研究建議 52
參考文獻 53

中文論文
王怡心，2016，善用「內控評估」強化專業判斷，會計研究月刊，第373期 (12月) : 56-60。
朱芯儀，2016，審計準則公報第58號「查核報告中關鍵查核事項之溝通」，會計研究月刊，第369期 (8月) : 88-92。
吳映儀，2018，關鍵查核事項與分析師預測行為之關聯性—基於台灣的實證研究，國立政治大學會計學系。
巫奕德，2018，經營風險及公司治理對券商分析師關注度之影響，銘傳大學財務金融學系碩士班。
巫鑫，2016，因應新式會計師查核報告－上市櫃公司辦理105年度決算最新實務，會計研究月刊，第373期 (12月) : 76-81。
李昭億，2019，在新式查核報告下，投資者對繼續經營存有疑慮之查核報告的反應，國立中正大學會計與資訊科技研究所。
周紹文，2016，探討文字指標對於企業績效的影響，國立中山大學資訊管理學系研究所。
徐永堅，2016，審計準則公報第五十八號「查核報告中關鍵查核事項之溝通」之內容、 影響與建議，證券暨期貨月刊，第34卷第9期 ( 9月) : 30-41。
馬秀如，2017，關鍵查核事項！？以T公司員工收賄為例，會計研究月刊，第375期 (2月) : 94-100。
張竣維，2018，以文字探勘法分析會計師簽證之查核報告，國立中正大學會計與資訊科技研究所。
陳威宇，2018，關鍵查核事項與審計品質之關聯性──兼論會計師事務所與會計師之影響，國立中正大學會計與資訊科技研究所。
陳穗婷、陳計良、陳虹吟，2019，新式查核報告對財務報表資訊揭露之影響，中原企管評論，第17卷第1期 ( 4月) : 59-82。
曾家璿、史雅男，2019，會計師產業專精與關鍵查核事項揭露之關聯，中華會計學刊，第15卷第1期 ( 4月) : 37-76。
湯玟珊，2013，盈餘管理行為與公司治理程度對信用評等之影響，國立高雄應用科技大學財富與稅務管理系。
黃娟娟，2012， 公司年報文字探勘與財務預警資訊內涵，逢甲大學商學博士學位學程。
葉乃菁、王玳琪、張嘉珍、吳騏、賴志遠，2009，建構創新政策研究工具文字探勘之應用簡介，國研科技，第22期 ( 4月) : 17-20。
廖政豪，2018，會計師查核報告中關鍵查核事項市場反應之研究，國立雲林科技大學會計系。
劉宛青，2016，巨量資料為基之集團企業財務報表舞弊偵測方法發展，國立高雄第一科技大學會計資訊研究所。
歐進士、李佳玲、詹茂昆，2004，我國企業盈餘管理與經營風險關聯之實證研究，風險管理學報，第6卷第2期 ( 7月) : 181-205。
謝松文，2010，以文字探勘技術萃取財務報告書之附註說明，國立中正大學會計與資訊科技研究所。
羅幼筎，2019，關鍵查核事項對負債資金成本的影響，國立中正大學會計資訊與法律數位學習碩士在職專班。

英文論文
Adnan Aziz, M., & Dar Humayon, A. 2006. Predicting corporate bankruptcy: where we stand? Corporate Governance: The international journal of business in society, 6(1), 18-33. doi:10.1108/14720700610649436
Altman, E. I. 1968. Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The Journal of Finance, 23(4), 589-609.
Alves Júnior, E. D., & Caio Galdi, F. 2020. The informational relevance of key audit matters. Relevância informacional dos principais assuntos de auditoria., 31(82), 67-83. doi:10.1590/1808-057x201908910
Aqel, S. 2014. Accounting and auditing independent auditor’s report: messages conveyed and implications. Acta Universitatis Danubius. Œconomica, 10(5), 5-17.
Azami, Z., & Salehi, T. 2017. The relationship between audit report delay and investment opportunities. Eurasian Business Review, 7(3), 437-449. doi:10.1007/s40821-016-0070-4
Beaver, W. H. 1966. Financial ratios as predictors of failure. Journal of Accounting Research, 4, 71-111. doi:10.2307/2490171
Burgstahler, D., & Dichev, I. 1997. Earnings management to avoid earnings decreases and losses. Journal of Accounting & Economics, 24(1), 99-126.
doi:https://doi.org/10.1016/S0165-4101(97)00017-7
Cai, S., Yuan, H., Lv, J., & Cui, Y. 2013. Application of IWO-SVM approach in fault diagnosis of analog circuits. 2013 25th Chinese Control and Decision Conference (CCDC), 4786-4791.
Camps-Valls, G., & Bruzzone, L. 2005. Kernel-based methods for hyperspectral image classification. IEEE Transactions on Geoscience and Remote Sensing, 43(6), 1351-1362.
Cecchini, M., Aytug, H., Koehler, G. J., & Pathak, P. 2010. Making words work: using financial text as a predictor of financial events. Decision Support Systems, 50(1), 164-175.
Chang, C.-C., & Lin, C.-J. 2011. LIBSVM: A library for support vector machines. ACM Transactions on Intelligent Systems and Technology (TIST), 2(3), 27:21-27:27.
Chen, P. F., He, S., Ma, Z., & Stice, D. 2016. The information role of audit opinions in debt contracting.Journal of Accounting & Economics, 61(1), 121-144.
Chen, Y.-J., Liou, W.-C., Chen, Y.-M., & Wu, J.-H. 2019. Fraud detection for financial statements of business groups. International Journal of Accounting Information Systems, 32, 1-23. doi:https://doi.org/10.1016/j.accinf.2018.11.004
Chen, Y.-J., Wu, C.-H., Chen, Y.-M., Li, H.-Y., & Chen, H.-K. 2017. Enhancement of fraud detection for narratives in annual reports. International Journal of Accounting Information Systems, 26, 32-45. doi:https://doi.org/10.1016/j.accinf.2017.06.004
Chen, Y.-J., & Wu, C.-Y. 2020. Predicting Corporate Financial Crisis using Letters to Shareholders, Soft Computing, In 2nd Review.
Christensen, B. E., Glover, S. M., & Wolfe, C. J. 2014. Do critical audit matter paragraphs in the audit report change nonprofessional investors' decision to invest? AUDITING: A Journal of Practice & Theory, 33(4), 71-93. doi:10.2308/ajpt-50793
Church, B. K., Davis, S. M., & McCracken, S. A. 2008. The auditor's reporting model: a literature overview and research synthesis. Accounting Horizons, 22(1), 69-90. doi:10.2308/acch.2008.22.1.69
Cortes, C., & Vapnik, V. 1995. Support-vector networks. Machine Learning, 20(3), 273-297. doi:10.1007/BF00994018
Dhaliwal, D. S. 1988. The effect of the firm's business risk on the choice of accounting methods. Journal of Business Finance & Accounting, 15(2), 289-302.
Ding, S., Yu, J., Qi, B., & Huang, H. 2014. An overview on twin support vector machines. Artificial Intelligence Review, 42(2), 245-252. doi:10.1007/s10462-012-9336-0
Fauvel, M., Chanussot, J., & Benediktsson, J. A. 2006. Evaluation of Kernels for Multiclass Classification of Hyperspectral Remote Sensing Data. Paper presented at the 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.
Fayyad, U. M., Piatetsky-Shapiro, G., Smyth, P., & Uthurusamy, R. 1996. Advances in knowledge discovery and data mining.
Feldman, R., & Sanger, J. 2007. The text mining handbook: advanced approaches in analyzing unstructured data. Cambridge university press.
Fuming, L., & Guo, J. 2011. Improving support vector machine by preprocessing data with decision tree. Paper presented at the 2011 International Conference on Computer Science and Service System (CSSS).
Gimbar, C., Hansen, B., & Ozlanski, M. E. 2016. Early evidence on the effects of critical audit matters on auditor liability. Current Issues in Auditing, 10(1), A24-A33.
Glosten, L. R., & Milgrom, P. R. 1985. Bid, ask and transaction prices in a specialist market with heterogeneously informed traders. Journal of Financial Economics, 14(1), 71-100. doi:https://doi.org/10.1016/0304-405X(85)90044-3
Köhler, A., Ratzinger-Sakel, N., & Theis, J. 2020. The effects of key audit matters on the auditor’s report’s communicative value: experimental evidence from investment professionals and non-professional investors. Accounting in Europe, 1-24. doi:10.1080/17449480.2020.1726420
KalipÇI ÇAĞIran, F., & Varici, İ. 2019. Key audit matters within the framework of ISA 701: an analysis on audit reports of companies listed in bist manufacturing industry. International Journal of Economics & Administrative Studies, 22, 193-208. doi:10.18092/ulikidince.467673
Kohavi, R. 1995. A study of cross-validation and bootstrap for accuracy estimation and model selection. Paper presented at the Ijcai.
Kulalı, İ. 2016. Altman z-score modelinin bist şirketlerinin finansal başarisizlik riskinin tahmin edilmesinde uygulanmasi. International Journal of Management Economics and Business, 12(27), 283-291.
Kuo, B.-C., Ho, H.-H., Li, C.-H., Hung, C.-C., & Taur, J.-S. 2014. A kernel-based feature selection method for svm with rbf kernel for hyperspectral image classification. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(1), 317-326. doi:10.1109/JSTARS.2013.2262926
Li, F. 2006. Do stock market investors understand the risk sentiment of corporate annual reports? , 54. Available at SSRN: https://ssrn.com/abstract=898181.
Li, F. 2008. Annual report readability, current earnings, and earnings persistence. Journal of accounting & economics, 45(2), 221-247.
doi:https://doi.org/10.1016/j.jacceco.2008.02.003
Li, F. 2010. Textual analysis of corporate disclosures: a survey of the literature. . Journal of accounting literature, 29, 143-165.
Li, H., & Zhang, Y. 2009. An algorithm of soft fault diagnosis for analog circuit based on the optimized SVM by GA. 2009 9th International Conference on Electronic Measurement&Instruments, 4-4-1027.
Manning, C. D., Raghavan, P., & Schütze, H. 2008. Introduction to information retrieval: Cambridge university press.
Maria, M. M., & Diana, H. I. 2019. Can we trust accounting result and cash-flow in apreciating financial performance? Annals of Faculty of Economics, 1(1), 175-184.
Menon, K., & Williams, D. D. 2010. Investor reaction to going concern audit reports. The Accounting Review, 85(6), 2075-2105. doi:10.2308/accr.2010.85.6.2075
Mock, T. J., Bédard, J., Coram, P. J., Davis, S. M., Espahbodi, R., & Warne, R. C. 2013. The audit reporting model: current research synthesis and implications. Auditing: A Journal of Practice & Theory, 32(Supplement 1), 323-351.
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., . . . Dubourg, V. (2011). Scikit-learn: Machine learning in Python. The Journal of Machine Learning Research, 12, 2825-2830.
Rafique, Y. Q., & Al, M. A. 2015. Corporate failure prediction of public listed companies in malaysia. European Researcher, 91(2), 114-126.
Schauerte, T., & Lindblad, F. 2015. Corporate economic distress in thewood construction industry: current state and trend after the economic crisis. Pro Ligno, 11(4), 389-396.
Sirois, L.-P., Bédard, J., & Bera, P. 2018. The informational value of key audit matters in the auditor's report: evidence from an eye-tracking study. Accounting Horizons, 32(2), 141-162.
Tan, A.-H. 1999. Text mining: The state of the art and the challenges. Paper presented at the Proceedings of the PAKDD 1999 Workshop on Knowledge Disocovery from Advanced Databases.
Tetlock, P. C. 2007. Giving content to investor sentiment: the role of media in the stock market. The Journal of Finance, 62(3), 1139-1168.
Vanstraelen, A., Schelleman, C., Meuwissen, R., & Hofmann, I. 2012. The audit reporting debate: seemingly intractable problems and feasible solutions. European accounting review, 21(2), 193-215. doi:10.1080/09638180.2012.687506
Xue, Z., Du, P., & Su, H. 2014. Harmonic analysis for hyperspectral image classification integrated with PSO optimized SVM. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(6), 2131-2146.
Yahav, I., Shehory, O., & Schwartz, D. 2019. Comments mining with TF-IDF: the inherent bias and its removal. IEEE Transactions on Knowledge and Data Engineering, 31(3), 437-450. doi:10.1109/TKDE.2018.2840127