Analysing trends of CSR implementation in Vietnamese banks: a deep learning-based approach

Organisations are increasingly adopting CSR initiatives to address social and environmental concerns while maintaining their profitability. However, analysing the vast amount of text data on CSR activities can be challenging and time-consuming. This study introduces an innovative method leveraging deep neural networks-based techniques to automatically detect CSR implementation topics within the annual reports, with a specific focus on the Vietnamese banking sector. We also contribute a manually annotated dataset labelled at two detailed levels using six main types of CSR topics and their corresponding 46 sub-topics to facilitate model development. The method's effectiveness is validated through comprehensive experiments on this dataset. Subsequently, based on the outcome, we perform further analysis to reveal the current status of CSR implementation and illuminate emerging trends of the 30 largest banks in Vietnam. This method helps organisations in automating trend detection, stakeholder engagement, competitor analysis, and reporting enhancement for strengthened CSR strategies.