Abstract: ［Purpose/Significance］This paper intends to explore the current development trend of image sentiment research, construct a research framework, theoretically promote the integration of image sentiment research and new methods, and provide theoretical reference and guidance for the follow-up research.［Design/Methodology］The VOSviewer was used to visually analyze the image sentiment research, and reveal the development trend of current related research. On this basis, the existing research was combed and summarized, and the corresponding research framework was developed.［Findings/Conclusion］Combining users' visual attention with image sentiment is an important trend of current research on image sentiment. Users’ visual attention is mostly presented by capturing their eye movement. A theoretical framework of imagesentiment research based on visual attention consists of data layer, method layer, content layer and application layer.［Originality/Value］The framework represents the characteristics of current research intuitively, theoretically promotes the combination of users’ eye movement and image sentiment, and provides foundation and guidance for the follow-up studies.

Keywords: Image sentiment, Visual attention, Visual sentiment analysis, Research framework