在氣候變遷加劇和淨零碳排放的促使下，環境永續議題儼然成為全球企業關注的焦點，尤其是從原料、製程到產品皆是環境永續發展重點的鋼鐵工業。為了深入探討鋼鐵業過去與現在對於環境永續所付出的努力，本研究運用文字探勘Guided LDA主題模型分析技術，結合依使用者設定特定主題的種子字進行模型訓練，將鋼鐵業企業社會責任報告書中所揭露的大量文字資訊，依照GRI準則的八項環境相關主題，歸納出主題在文件中的機率分布和單詞在主題中的機率分布。最後我們透過隨機森林分類模型和皮爾森相關係數矩陣，依照年度、MSCI ESG等級、企業所在地等三大構面，詮釋實驗結果中鋼鐵業之企業社會責任報告書所隱含的環境永續發展相關趨勢和特性。我們發現多數企業在五年內的環境主題分布非常相似，唯獨受到GRI準則改版的些微影響；而企業所在地則是影響環境主題分布的主要因素，間接證明國家政策和法規對環境永續的影響力。

Due to the urgence of rapid climate change and zero carbon emission policy, environmental sustainability issues become much more significant in every industry throughout the world, especially steel industry which is highlighted in environmental sustainability from materials to products. In order to understand the past and present efforts of the steel industry towards environmental sustainability, we utilize topic modeling technique called Guided LDA model, which combines several user-defined seed word sets in specific topics for model training. According to eight environmental related topics of GRI standard, a large amount of text information of CSR report could be analyzed by Guided LDA model, via topic-document distribution and word-topic distribution. After getting results from random forest regression model and Pearson correlation matrix, we explain the experimental results from three factors including year, MSCI ESG rating and location of company. And the results show environmental sustainability trends and characteristics in CSR report in steel industry. We found that the distribution of environmental topics over a five-year period was very similar for most companies, with only a slight impact from the revision of the GRI index. The location of companies is the main factor affecting the distribution of environmental topics, which indirectly demonstrates the influence of national policies and regulations on environmental sustainability.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
目錄 v
圖次 vii
表次 ix
第一章、緒論 1
第一節、研究背景與動機 1
第二節、研究目的 2
第三節、章節安排 3
第二章、文獻探討 4
第一節、企業社會責任與環境永續議題 4
第二節、文字探勘技術應用於企業社會責任 6
第三節、主題模型 8
第三章、研究方法 12
第一節、研究架構 12
第二節、資料蒐集 13
第三節、資料前處理 18
第四節、Guided LDA模型 19
第五節、Pearson相關係數 22
第四章、實驗結果與討論 23
第一節、主題模型分析結果 23
第二節、主題機率分布相關性分析 31
第五章、結論與未來展望 45
參考文獻 46
附　錄 49

Aureli, S. (2017). A comparison of content analysis usage and text mining in CSR corporate disclosure. International Journal of Digital Accounting Research, 17.
Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. Journal of machine Learning research, 3(Jan), 993-1022.
Butlin, J. (1989). Our common future. By World commission on environment and development.(London, Oxford University Press, 1987, pp. 383£ 5.95.).
CSRone永續智庫, 【第七屆發表會】臺灣永續報告首度突破600本 透明揭露ESG資訊, 2021, https://csrone.com/topics/6811, [Retrieved 2022/05].
De George, R. T. (1987). The status of business ethics: Past and future. Journal of Business ethics, 6(3), 201-211.
Desa, U. N. (2016). Transforming our world: The 2030 agenda for sustainable development.
Giese, G., Lee, L. E., Melas, D., Nagy, Z., & Nishikawa, L. (2019). Foundations of ESG investing: How ESG affects equity valuation, risk, and performance. The Journal of Portfolio Management, 45(5), 69-83.
Global Reporting, GRI Standards Traditional Chinese Translations, 2022, https://www.globalreporting.org/how-to-use-the-gri-standards/gri-standards-traditional-chinese-translations/, [Retrieved 2022/06].
Global Reporting Initiative. (2016). GRI 101: Foundation 2016. Amsterdam: GRI.
Goloshchapova, I., Poon, S. H., Pritchard, M., & Reed, P. (2019). Corporate social responsibility reports: topic analysis and big data approach. The European Journal of Finance, 25(17), 1637-1654.
Guetterman, T. C., Chang, T., DeJonckheere, M., Basu, T., Scruggs, E., & Vydiswaran, V. V. (2018). Augmenting qualitative text analysis with natural language processing: methodological study. Journal of medical Internet research, 20(6), e9702.
Hardeniya, N., Perkins, J., Chopra, D., Joshi, N., & Mathur, I. (2016). Natural language processing: python and NLTK. Packt Publishing Ltd.
Jagarlamudi, J., Daumé III, H., & Udupa, R. (2012, April). Incorporating lexical priors into topic models. In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics (pp. 204-213).
Jaworska, S., & Nanda, A. (2018). Doing well by talking good: A topic modelling-assisted discourse study of corporate social responsibility. Applied Linguistics, 39(3), 373-399.
Kiriu, T., & Nozaki, M. (2020). A Text Mining Model to Evaluate Firms’ ESG Activities: An Application for Japanese Firms. Asia-Pacific Financial Markets, 27(4), 621-632.
Maier, D., Waldherr, A., Miltner, P., Wiedemann, G., Niekler, A., Keinert, A., ... & Adam, S. (2018). Applying LDA topic modeling in communication research: Toward a valid and reliable methodology. Communication Methods and Measures, 12(2-3), 93-118.
Michelon, G., Pilonato, S., & Ricceri, F. (2015). CSR reporting practices and the quality of disclosure: An empirical analysis. Critical perspectives on accounting, 33, 59-78.
Moravcikova, K., Stefanikova, Ľ., & Rypakova, M. (2015). CSR reporting as an important tool of CSR communication. Procedia Economics and finance, 26, 332-338.
MSCI, ESG Rating, 2022, https://www.msci.com/our-solutions/esg-investing/esg-ratings, [Retrieved 2022/06].
MSCI, ESG Industry Materiality Map, 2022, https://www.msci.com/our-solutions/esg-investing/esg-industry-materiality-map#, [Retrieved 2022/06].
Park, K., & Kremer, G. E. O. (2017). Text mining-based categorization and user perspective analysis of environmental sustainability indicators for manufacturing and service systems. Ecological Indicators, 72, 803-820.
Pisani, N., Kourula, A., Kolk, A., & Meijer, R. (2017). How global is international CSR research? Insights and recommendations from a systematic review. Journal of World Business, 52(5), 591-614.
Plisson, J., Lavrac, N., & Mladenic, D. (2004, October). A rule based approach to word lemmatization. In Proceedings of IS (Vol. 3, pp. 83-86).
Singh V. (2017), Welcome to GuidedLDA’s documentation, https://guidedlda.readthedocs.io/en/latest/, [Retrieved 2022/06].
Slacik, J., & Greiling, D. (2020). Coverage of G4-indicators in GRI-sustainability reports by electric utilities. Journal of Public Budgeting, Accounting & Financial Management.
Smeuninx, N., De Clerck, B., & Aerts, W. (2020). Measuring the readability of sustainability reports: A corpus-based analysis through standard formulae and NLP. International Journal of Business Communication, 57(1), 52-85.
Srividhya, V., & Anitha, R. (2010). Evaluating preprocessing techniques in text categorization. International journal of computer science and application, 47(11), 49-51.
Te Liew, W., Adhitya, A., & Srinivasan, R. (2014). Sustainability trends in the process industries: A text mining-based analysis. Computers in Industry, 65(3), 393-400.
Vijayarani, S., Ilamathi, M. J., & Nithya, M. (2015). Preprocessing techniques for text mining-an overview. International Journal of Computer Science & Communication Networks, 5(1), 7-16.
Zhou, H., Deng, Z., Xia, Y., & Fu, M. (2016). A new sampling method in particle filter based on Pearson correlation coefficient. Neurocomputing, 216, 208-215.