<p>In the digital era, corporate social responsibility (CSR) has become a crucial factor influencing consumer purchase behavior. To investigate the mechanism through which CSR affects customer purchase intention (CPI) and to examine the mediating role of Brand Equity (BE), this study integrates deep learning techniques with social identity theory to construct a comprehensive research framework supported by multi-source data. A two-stage approach is employed. In the first stage, an optimized Text-Convolutional Neural Network model is applied to perform sentiment analysis on consumer reviews from the publicly available Yelp dataset, encompassing sectors such as food services, hospitality, and beauty. This model effectively captures consumer sentiment toward CSR-related initiatives. In the second stage, a structured questionnaire is developed based on social identity theory to quantitatively examine the relationships among CSR, BE, and CPI. The empirical results reveal that CSR exerts a significant positive influence on CPI and that BE plays a partial mediating role in this relationship. These findings suggest that CSR activities enhance purchase intention by fostering stronger brand identification. Notably, CSR efforts related to environmental protection and public welfare demonstrate a particularly strong influence on consumer behavior. Theoretically, this study extends the application of social identity theory in the CSR domain by introducing an integrated framework that combines sentiment recognition with structural equation modeling. Methodologically, the study advances a cross-disciplinary approach by merging natural language processing with survey-based research to facilitate the deep integration of multi-source data. Practically, this study provides data-driven evidence and strategic insights for enterprises aiming to design effective CSR strategies and enhance brand management, offering valuable implications for managerial decision-making.</p>