Cyberspace has significantly influenced people’s perceptions of social interactions and communication. As a result, the conventional theories of kin selection and reciprocal altruism fall short in completely elucidating online prosocial behavior. Based on the social information processing model, we propose an analytical framework to explain the donation behaviors on online platform. Through collecting textual and visual data from Tencent Gongyi platform pertaining to disease relief projects, and employing techniques encompassing text analysis, image analysis, and propensity score matching, we investigate the impact of both internal emotional cues and external contextual cues on donation behaviors. It is found that positive emotions tend to attract a larger number of donations, while negative emotions tend to result in higher per capita donation amounts. Furthermore, these effects manifest differently under distinct external contextual conditions.