Corporate social responsibility (CSR) has gained a great deal of interest in recent years due to the need for information that can help many stakeholders (e.g., governments, investors, professional organizations, researchers, etc.) understand companies’ contributions to the environment and society. CSR disclosure (CSRD) is now the key source of such information when analyzing, for example, an institution’s future performance. In the current body of CSRD literature, the majority of quantitative CSRD studies have relied on traditional statistical approaches for the correlation analysis of CSRD influencing factors. In this paper, we intend to quantitatively analyze firms’ characteristics related to CSRD in Saudi Arabia, understand CSRD and its influencing factors, and predict CSRD patterns. This study lays the groundwork to help companies make informed decisions. It also helps many other stakeholders better understand CSRD’s impacts. To achieve this, we propose a deep learning framework based on long short-term memory (LSTM) for identifying and predicting CSRD patterns. Moreover, a correlation-based technique is also used to visualize the relationships between variables and identify the significant features. The dataset used in this study was collected from annual reports, CSR reports, and firms’ websites between 2015 and 2018. It contains a variety of variables to explain the CSR behaviour of 117 companies. The proposed framework is evaluated with several approaches, including logistic regression (LR), K-nearest neighbours (KNN), support vector machines (SVM), random forests (RF), and decision trees (DT). Compared to other machine learning models, experiment results show that LSTM achieved acceptable results with the highest accuracy of \(88_\%\).

近年来，由于需要能够帮助许多利益相关者（例如政府、投资者、专业组织、研究人员等）了解公司对环境和环境的贡献的信息，企业社会责任（CSR）引起了极大的关注。社会。企业社会责任披露（CSRD）现在是分析机构未来绩效等信息的主要来源。在目前的CSRD文献中，大多数定量CSRD研究都依赖于传统的统计方法来进行CSRD影响因素的相关性分析。在本文中，我们打算定量分析沙特阿拉伯企业与CSRD相关的特征，了解CSRD及其影响因素，并预测CSRD模式。这项研究为帮助公司做出明智的决策奠定了基础。它还帮助许多其他利益相关者更好地了解 CSRD 的影响。为了实现这一目标，我们提出了一种基于长短期记忆（LSTM）的深度学习框架，用于识别和预测 CSRD 模式。此外，还使用基于相关性的技术来可视化变量之间的关系并识别重要特征。本研究使用的数据集来自2015年至2018年的年度报告、企业社会责任报告和企业网站。它包含多种变量来解释117家企业的企业社会责任行为。所提出的框架使用多种方法进行评估，包括逻辑回归（LR）、K近邻（KNN）、支持向量机（SVM）、随机森林（RF）和决策树（DT）。与其他机器学习模型相比，实验结果表明 LSTM 取得了可以接受的结果，最高准确率达到\(88_\%\)。