VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL GRADUATION PROJECT PROJECT NAME Analyzing trends in implementing Corporate Social Responsibility using NLP tools Student’s name Nguyễn Mạnh Tiến Hanoi - Year 2023 VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL GRADUATION PROJECT PROJECT NAME Analyzing trends in implementing Corporate Social Responsibility using NLP tools SUPERVISOR: Associate Prof. Tran Thi Oanh STUDENT: Nguyễn Mạnh Tiến STUDENT ID: 19071006 COHORT: Computer Science SUBJECT CODE: INS401101 MAJOR: Business Data Analytics Hanoi - Year 2023 Acknowledgement Throughout the span of four months dedicated to my graduation project, which was aligned with the objectives set forth by the International School - Hanoi National University, I had the opportunity to delve into the fascinating realm of analyzing trends in implementing corporate social responsibility (CSR) within the banking industry in Vietnam. I came to recognize that the foundational knowledge acquired through my university coursework played a pivotal role in enabling me to engage in data analytics work across diverse business domains. As I embarked on this project, I not only applied the knowledge acquired during my studies but also delved into extensive research on the Natural Language Processing field, striving to amalgamate both domains.
This endeavor has been immensely enlightening, serving as the most significant takeaway from my graduation project. In expressing my sincere gratitude and utmost respect, I would like to extend my heartfelt appreciation to my esteemed instructor, Associate Prof. Tran Thi Oanh. She has been an unwavering source of guidance, supporting me right from the outset of my academic journey with foundational subjects such as database systems, and continuing to provide invaluable guidance throughout the completion of my graduation project.
Despite encountering numerous challenges throughout this undertaking, with the invaluable assistance of my instructor, I was able to overcome obstacles and ultimately bring the project to fruition. I also want to thank PhD. Nguyen Phuong Mai for supporting the business domain about CSR, Chu Minh Thanh, Nguyen Thi Thu, and Nguyen Thi Huyen for supporting me in labeling text data. Thanks everyone sincerely! 1 Abstract This study aims to analyze the trends in implementing corporate social responsibility (CSR) in the banking industry using Natural Language Processing (NLP) tools.
With the growing emphasis on sustainability and ethical business practices, CSR has gained significant importance in recent years. Organizations are increasingly adopting CSR initiatives to address social and environmental concerns while maintaining their profitability. However, analyzing and understanding the vast amount of textual data related to CSR activities can be challenging and time-consuming. In this study, we propose the use of NLP tools to analyze textual data (annual reports) to identify and quantify CSR trends.
NLP techniques enable the extraction of valuable insights from unstructured textual data, allowing for a comprehensive understanding of CSR implementation across industries and organizations. 2 Table of Contents Acknowledgement. 2 List of Figures. 5 List of Tables.
The necessity of the project. About Corporate Social Responsibility (CSR). Applied Artificial Intelligence in CSR. Aims and Objectives.
11 Chapter 2: Theoretical Background. Basic Concept of NLP algorithms. 13 Chapter 3: Proposed Architecture for Classifying CSR sub-topics using NLP tools. Building NLP Models & Evaluation.
23 Chapter 4: Experiments and Discussion. About the Dataset. About the labels. Exploratory Data Analysis (EDA).
Transform PDF files into text. Standardize text structure. Preparing input for NLP models. Build NLP models.
Problem 1: CSR or No CSR. Problem 2: CSR topics. Problem 3: CSR sub-topics. Final NLP model.
44 Chapter 5: Conclusion & Further Experiments. Limitations & Further Experiments. 47 4 List of Figures Figure 1: Visualization for SVM. 14 Figure 2: Random Forest Simplified.
15 Figure 3: ANN architecture. 16 Figure 4: Fine-tuning classifier architecture of BERT in classification. 18 Figure 5: Overall Workflow. 20 Figure 6: Data Collection Workflow.
21 Figure 7: Data Preprocessing Workflow. 22 Figure 8: Model Building & Evaluation Workflow. 23 Figure 9: Image-only PDF example - LienVietPostBank 2017 Annual Report. 24 Figure 10: Digitally created PDF example - VPBank 2022 Annual Report.
24 Figure 11: Dataset used in this project. 25 Figure 12: Interpretation of Cohen's Kappa. 27 Figure 13: Number of CSR activities throughout the years. 30 Figure 14: CSR activities' proportion throughout the years.
31 Figure 15: Details for CSR to employees. 32 Figure 16: Details for CSR to the community. 33 Figure 17: Top 10 most active banks in CSR in Vietnam. 34 Figure 18: Types of PDF files.
35 Figure 19: Example on how OCR detect text from Image-Only PDF. 35 Figure 20: Transformed Dataset. 37 Figure 21: Example of Kappa value between annotators 3 and 4. 37 Figure 22: Top 5 most frequently appear stopwords.
40 Figure 23: Code to tokenize sentences. 41 Figure 24: Code to create attention masks. 41 Figure 25: Classification Report of our workflow on test data. 45 5 List of Tables Table 1: List of CSR topics.
27 Table 2: Cohen's Kappa values between annotators. 37 Table 3: Annotators' Contribution. 38 Table 4: Performance on the test set of Problem 1. 43 Table 5: Performance on the test set of Problem 2.
43 Table 6: Best performance model for each topic. 44 6 Abbreviations & Acronyms CSR Corporate Social Responsibility NLP Natural Language Processing AI Artificial Intelligence ML Machine Learning DL Deep Learning SVM Support Vector Machine CNN Convolutional Neural Networks LSTM Long Short-Term Memory BERT Bidirectional Encoder Representations from Transformers NER Named Entity Recognition NLI Natural Language Inference EDA Exploratory Data Analysis GRI Global Reporting Initiative 7 Chapter 1: Introduction 1. Problem Statement In the context of utilizing Natural Language Processing (NLP) tools, there is a need to analyze the trends in Corporate Social Responsibility (CSR) practices adopted by businesses. The problem at hand is to identify and assess the extent to which businesses are integrating social responsibility into their operations using NLP-based analysis.
By understanding the current landscape of CSR practices and their alignment with NLP techniques, we can address the challenges and opportunities in effectively leveraging NLP for monitoring and evaluating corporate social responsibility efforts. The necessity of the project Analyzing trends in implementing corporate social responsibility (CSR) using natural language processing (NLP) tools is a project that holds significant necessity in today's corporate landscape. By harnessing the power of NLP, corporations can effectively detect and analyze CSR topics, leading to a myriad of benefits for their operations and reputation. Firstly, NLP tools offer corporations the ability to efficiently and accurately sift through vast amounts of textual data from diverse sources, such as social media platforms, news articles, and customer feedback.
This enables them to identify emerging CSR trends and concerns in real-time, allowing for proactive decision-making and strategic planning. By staying ahead of the curve, companies can address societal issues promptly and tailor their CSR initiatives to meet the evolving needs of their stakeholders. Secondly, NLP tools provide a means to gain deeper insights into public sentiment towards CSR initiatives. By analyzing sentiments expressed in online conversations and media coverage, corporations can gauge the effectiveness of their existing CSR programs and make data-driven adjustments when necessary.
This valuable feedback loop enables companies to align their CSR efforts with the expectations and values of their customers, employees, and other stakeholders, resulting in increased trust and loyalty. Additionally, NLP tools enable corporations to monitor and assess their competitors' CSR activities. By analyzing publicly available information, such as annual reports, press releases, and sustainability disclosures, companies can benchmark their CSR performance against industry peers. This competitive analysis allows corporations to identify areas where they can 8 improve, differentiate themselves, and develop unique CSR strategies that align with their organizational values and strengths.
Lastly, the use of NLP tools in analyzing CSR trends provides corporations with valuable insights for reporting and communication purposes. Accurate and comprehensive CSR reporting is essential for maintaining transparency and accountability. NLP tools can automate the extraction of CSR-related information from various sources, making the reporting process more efficient and accurate. Additionally, by understanding the prevalent CSR topics and concerns in public discourse, companies can communicate their CSR initiatives more effectively, addressing the issues that matter most to their stakeholders and enhancing their overall corporate reputation.
In conclusion, the project of analyzing trends in implementing corporate social responsibility (CSR) using NLP tools is highly necessary for corporations seeking to navigate the complex landscape of social responsibility. By leveraging NLP capabilities, companies can benefit from real-time trend detection, enhanced stakeholder engagement, competitor analysis, and improved reporting and communication. Ultimately, these advantages translate into a more robust and effective CSR strategy, fostering positive societal impact while enhancing corporate reputation and long-term sustainability. About Corporate Social Responsibility (CSR) Before going to details, let’s first introduce Corporate Social Responsibility (CSR).
According to Zoe Hansen, CSR is a self-regulating business model that helps a company be socially accountable to itself, its stakeholders, and the public.[1] Through these activities, companies can improve their aspects of society and promote a positive brand image for them. The CSR is taken place in 6 main areas, such as plant closures, employee relations, human rights, corporate ethics, community relations and the environment. [2] There are many benefits that a company can get from displaying CSR activities. It is shown in many studies that even though CSR might entail short-term costs, it will bring many advantages for companies in the long run.
Enhanced Reputation and Brand Image: By actively engaging in CSR initiatives, companies can build a positive reputation and enhance their brand image. Consumers and stakeholders are more likely to trust and support companies that demonstrate a commitment to social and environmental issues. A strong reputation can lead to increased customer loyalty, improved market position, and a competitive advantage. Stakeholder Engagement and Relations: CSR helps foster strong relationships with stakeholders, including employees, customers, suppliers, investors, and local communities.
By addressing their concerns and contributing to their well-being, companies can improve stakeholder engagement and create a sense of loyalty and trust. This can lead to enhanced employee morale, customer satisfaction, and stronger partnerships with suppliers and investors. Risk Management and Regulatory Compliance: CSR initiatives enable companies to proactively manage risks associated with social and environmental issues. By adopting responsible practices and adhering to regulations, companies can minimize legal and reputational risks.
This includes mitigating the potential negative impacts of non- compliance, such as fines, lawsuits, and damage to brand reputation. Innovation and Competitive Advantage: Embracing CSR can drive innovation within organizations. Companies that actively seek sustainable and socially responsible solutions can uncover new business opportunities and develop innovative products and services. This can provide a competitive advantage by differentiating them from competitors and attracting environmentally and socially conscious customers.
Employee Recruitment and Retention: CSR initiatives contribute to attracting and retaining talented employees. Today's workforce values purpose-driven work environments and seeks to align their personal values with their employers' values. Companies that prioritize CSR initiatives can attract top talent, enhance employee satisfaction, and reduce turnover rates. Cost Savings and Efficiency: Implementing CSR practices can lead to cost savings and operational efficiencies.
For example, adopting energy-efficient technologies and sustainable supply chain practices can reduce resource consumption and waste, resulting in lower operational costs. CSR initiatives that focus on employee well-being and development can also lead to increased productivity and reduced absenteeism. In summary, CSR brings numerous benefits to companies, including improved reputation, stakeholder engagement, risk management, innovation, employee recruitment and retention, and cost savings. Embracing CSR is not only socially responsible but also strategically advantageous for long-term business success.
Applied Artificial Intelligence in CSR In this era of booming technological innovations, Artificial Intelligence (AI) methods are used to analyze and compare sustainability-related textual data, which will increase efficiency 10 and its scalability. A content analysis of texts will provide valuable insights, such as upcoming CSR trends. It also allows sectoral benchmarking or quality checks for sustainability reporting.