近年來，在公司治理及永續發展的浪潮下，企業投入心力對社會做出貢獻，以報告書的方式揭露公司在社會責任的目標、成果、承諾及規劃。報告書參照多項GRI 指標撰寫，揭露公司對經濟、環境及社會的影響，是CSR評鑑的重要標準。為了自動快速預測文本所參照的指標，本論文篩選各項環境GRI指標的種子字，利用GuidedLDA探測文本揭露的環境指標項目 。我們蒐集了台灣上市公司的CSR報告書，擷取參照GRI指標撰寫的文章段落，在實驗中發現，小樣本的情況下，我們的方法分辨文本揭露的環境指標項目，準確度上優於現有的分類器，且有能力判斷是否參照GRI環境指標。

Corporate governance and sustainable development have been taken seriously by many enterprises. Enterprise strives to contribute to society, and they publish Corporate Social Responsibility reports every year to record their goals, results, commitment and planning. According to the GRI Standard, CSR reports describe the impact of companies to economical, environmental and social aspects. There are several criteria for each aspect that can be used for CSR report evaluation. It is essential that a CSR report meets each of the criteria. In this thesis, we propose an approach to automatically predict whether a given CS report will meet each criterion in the environmental aspect. Our approach applies GuidedLDA for such a task. We collected the CSR reports from listed companies of Taiwan and captured the paragraphs that have the potential meeting the GRI Standard. We found from experiments that, in small data set our approach performs better than all other traditional learning methods. For larger data set, our approach outperform unsupervised learning approach and has comparable performance with the supervised learning approach yet being able to provide interpretations.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
Table Of Content vi
Table Of Figure vii
Table Of Table viii
CHAPTER 1 – Introduction 1
CHAPTER 2 – Related Work 5
1. Corporate social responsibility 5
2. Corpus Modeling 6
3. Text-mining in CSR reports 10
CHAPTER 3 – Methodology 14
1. Task description 14
2. Datasets collection 14
3. Our approach 19
CHAPTER 4 – Experimental Evaluation 31
1. Unsupervised Learning methods 31
2. Supervised Learning methods 37
3. The impact of the quality of the seed words on the method 39
4. Add irrelevant environmental indicator data in data sets 40
CHAPTER 5 – Conclusion 42
Reference 43

Bowen, H. (1953). Social responsibilities of the businessman. New York: Harper.
Sethi, S. P. (1975). Dimension of Corporate Social Performance: An Analytical Framework. California Management Review, 17(3), 58-64.
Carroll, A. B. (1979). A three-dimensional model of corporate performance. Academy of Management Review, 4, 495-505.
Goodpaster, K. (1991). Business ethics & stakeholder analysis. Business Ethics Quarterly, 1, 53-73.
Epstein E.M. (1989). Corporate good citizenship and the corporate social policy process : a view from the United State. Journal of Business Ethics, Vol.8 1989 pp.583-595
Maclagan P. W. (1998). Management and Morality (Sage Publications, London)
Hay. R.D., E.R. Gray and J.E. Gates (1976). Business and Society, Southwestern Publishing Cincinnati
Salton, G., & McGill, M. J. (1983). Introduction to modern information retrieval. mcgraw-hill.
Deerwester, S., Dumais, S. T., Furnas, G. W., Landauer, T. K., & Harshman, R. (1990). Indexing by latent semantic analysis. Journal of the American society for information science, 41(6), 391-407.
Hofmann, T. (1999, August). Probabilistic latent semantic indexing. In Proceedings of the 22nd annual international ACM SIGIR conference on Research and development in information retrieval (pp. 50-57).
Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. the Journal of machine Learning research, 3, 993-1022.
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).
Liu, S. H., Chen, S. Y., & Li, S. T. (2017, July). Text-mining application on CSR report analytics: A study of petrochemical industry. In 2017 6th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI) (pp. 76-81). IEEE.
Lee, M., & Huang, Y. L. (2020). Corporate social responsibility and corporate performance: A hybrid text mining algorithm. Sustainability, 12(8), 3075.
Song, Y., Wang, H., & Zhu, M. (2018). Sustainable strategy for corporate governance based on the sentiment analysis of financial reports with CSR. Financial Innovation, 4(1), 1-14.
Saha, A., & Nabareseh, S. (2015). Communicating corporate social responsibilities: Using text mining for a comparative analysis of banks in India and Ghana. Mediterranean Journal of Social Sciences, 6(3 S1), 11-11.
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.
Uekado, K., Feng, L., Suzuki, M., & Ohwada, H. (2018, August). Classification of CSR Using Latent Dirichlet Allocation and Analysis of the Relationship Between CSR and Corporate Value. In Pacific Rim Knowledge Acquisition Workshop (pp. 261-270). Springer, Cham.
Tremblay, M. C., Parra, C., & Castellanos, A. (2015, May). Analyzing corporate social responsibility reports using unsupervised and supervised text data mining. In International Conference on Design Science Research in Information Systems (pp. 439-446). Springer, Cham.
Shahi, A. M., Issac, B., & Modapothala, J. R. (2014). Automatic analysis of corporate sustainability reports and intelligent scoring. International Journal of Computational Intelligence and Applications, 13(01), 1450006.

Chen, T., & Guestrin, C. (2016, August). Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining (pp. 785-794).