隨者氣候的變遷，很多企業開始重視永續發展。企業希望能在最小化對環境 影響的條件下，提升公司的總體營運。不僅降低對地球的傷害，也可以提升社會 大眾對企業的信任。而對企業而言，曝光自身對永續發展貢獻的方式除了新聞之 外，就是永續報告書。近幾年也有很多這方面的研究，但目前還沒有針對台灣企 業的研究，因此我們想要探討台灣 ICT 這種對環境影響極大的產業，他的永續報 告書對企業的影響為何。
本研究的目的是透過文字探勘的技術來分析永續報告書與企業財務表現的關 係，藉此找到永續發展與企業財務表現的關聯性。
論文中使用的方法是透過現在的 SOTA 文字模型 DeBERTa 進行訓練，並根 據我們要進行的多標籤分類任務進行微調，皆遮對台灣 ICT 產業的永續報告書進 行標籤。
為了使我們的結果更具公信力，我們使用階層式分群法(Hierarchical Clustering)針對報告書標籤結果進行分群 ，在透過成分分析法(Principal Component Analysis)取出前三個主成分作為每一群的特徵系數，並以此特徵 值作為後續回歸分析，得出永續報告書與財務表現的影響顯不顯著。經過深入研 究後，我們發現永續報告書與台灣的永續發展指標關係並不顯著，但台灣的永續 發展指標卻與財務表現關係非常顯著。透過本論文的研究可以為企業確立永續發 展對他們的營運有正向的影響，幫助企業為總體營運的發展下更明確的決策。

In the context of climate change, many companies are now focusing on sustainable development. They aim to minimize their environmental impact while enhancing overall operational efficiency. This approach not only reduces harm to the planet but also boosts public trust. For corporations, showcasing their commitment to sustainability involves more than just press releases; it includes publishing detailed sustainability reports. Although there has been significant research in this field, studies specifically targeting Taiwanese companies are lacking. Thus, our research aims to examine the influence of sustainability reports on companies within Taiwan's ICT industry, known for its substantial environmental impact.
The objective of this study is to explore the relationship between sustainability reports and corporate financial performance using text mining techniques to identify correlations between sustainability efforts and financial outcomes.
Our methodology leverages the state-of-the-art text model, DeBERTa, which is trained and fine-tuned for multiple labeling classification tasks specific to sustainability reports of Taiwan’s ICT industry. To enhance the credibility of our findings, we employ Hierarchical Clustering to organize the report labeling results, and then extract the top three principal components as characteristic coefficients for each cluster through Principal Component Analysis. These coefficients are subsequently used as regression variables to analyze the impact of sustainability reports on financial performance. Our analysis indicates that the direct effect of sustainability reports on financial performance is not significant. However, a deeper investigation reveals that while the relationship between sustainability reports and Taiwan’s sustainability indicators is weak, the link between these sustainability indicators and financial performance is notably strong. This study confirms that sustainability positively influences corporate operations, which can
ii
assist businesses in making more informed decisions regarding their overall strategic direction.

論文審訂書 i
摘要 ii
ABSTRACT iii
CHAPTER 1 INTRODUCTION1
1.1 Research Background1
1.2 Research Motivation and Purpose3
1.3 Organization of the Research4
CHAPTER 2 LITERATURE REVIEW5
2.1 Sustainable development5
2.1.1 ESG Definition6
2.1.2 ESG and CSR8
2.1.3 ESG Impacts9
2.1.4 ESG Disclosure11
2.1.5 Data Mining13
2.2 BERT Model15
2.2.1 Encoder15
2.2.2 Transformer19
CHAPTER 3 METHODOLOGY21
3.1 Data Collecting and Preprocessing22
3.1.1 Labeled ESG Report Corpus23
3.1.2 ESG Agenda Setting Corpus26
3.1.3 Corporate Financial Indicators27
3.2 DeBERTa Fine-tuning28
3.3 Sustainability Report Processing29
3.3.1 Sentence-based Labeling29
3.3.2 Review-based Aggregating30
3.3.3 ESG Report Profiling31
3.4 ESG Performance Analyzing33
CHAPTER 4 RESULT35
4.1 Description of the Datasets35
4.2 Fine-tuned DeBERTa with Evaluation37
4.3 Clustering Analysis Results and Agenda Settings39
4.4 Regression Analysis44
4.4.1 Result I: TESG and Corporate Revenue44
4.4.2 Result II: Textual ESG Categories and TESG46
4.5 Managerial implications52
CHAPTER 5 CONCLUSION53
5.1 Conclusion Remarks53
5.2 Limitations and Future Work55
REFERENCES56

Alareeni, B. A., & Hamdan, A. (2020). ESG impact on the performance of US S&P 500-listed firms. Corporate Governance: The International Journal of Business in Society, 20(7), 1409–1428. https://doi.org/10.1108/CG-06-2020-0258
Cormier, D., & Magnan, M. (1999). Corporate Environmental Disclosure Strategies: Determinants, Costs and Benefits. Journal of Accounting, Auditing & Finance, 14(4), 429–451. https://doi.org/10.1177/0148558X9901400403
Darnall, N., Ji, H., Iwata, K., & Arimura, T. H. (2022). Do ESG reporting guidelines and verifications enhance firms’ information disclosure? Corporate Social Responsibility and Environmental Management, 29(5), 1214–1230. https://doi.org/10.1002/csr.2265
Fatemi, A., Glaum, M., & Kaiser, S. (2018). ESG performance and firm value: The moderating role of disclosure. Global Finance Journal, 38, 45–64. https://doi.org/10.1016/j.gfj.2017.03.001
Greiner, M., & Sun, J. (2021). How corporate social responsibility can incentivize top managers: A commitment to sustainability as an agency intervention. Corporate Social Responsibility and Environmental Management, 28(4), 1360–1375. https://doi.org/10.1002/csr.2148
Hasan, I., Kobeissi, N., Liu, L., & Wang, H. (2018). Corporate Social Responsibility and Firm Financial Performance: The Mediating Role of Productivity. Journal of Business Ethics, 149(3), 671–688. https://doi.org/10.1007/s10551-016-3066-1
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. https://doi.org/10.1007/s10690-020-09309-1
Liew, W. T., Adhitya, A., & Srinivasan, R. (2014). Sustainability trends in the process industries: A text mining-based analysis. Computers in Industry, 65(3), 393–400. https://doi.org/10.1016/j.compind.2014.01.004
Sethi, S. P., Martell, T. F., & Demir, M. (2017). Enhancing the Role and Effectiveness of Corporate Social Responsibility (CSR) Reports: The Missing Element of Content Verification and Integrity Assurance. Journal of Business Ethics, 144(1), 59–82. https://doi.org/10.1007/s10551-015-2862-3
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2021). An image is worth 16x16 words: Transformers for image recognition at scale. International Conference on Learning Representations (ICLR).
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems (NeurIPS), 30.
Peters, M. E., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018). Deep contextualized word representations. Computing Research Repository (CoRR), abs/1802.05365.
Bahdanau, D., Cho, K., & Bengio, Y. (2015). Neural machine translation by jointly learning to align and translate. Computing Research Repository (CoRR).
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. Association for Computational Linguistics (ACL).
Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2020). Roberta: A robustly optimized bert pretraining approach. International Conference on Learning Representations (ICLR).
Xie, D., Ji, J., Xu, J., & Ji, R. (2021). Combining improvements for exploiting dependency trees in neural semantic parsing. Pacific Rim International Conference on Artificial Intelligence (PRICAI), 58–72.
Arvidsson, S., & Dumay, J. (2022). Corporate ESG reporting quantity, quality and performance: Where to now for environmental policy and practice? Business Strategy and the Environment, 31(3), 1091-1110.