Multi-modal sentiment analysis method for cross-modal supplementary balance
Through the multimodal sentiment analysis method of cross-modal complementary balance, the correlation and difference between modalities are enhanced, the importance of prediction results is dynamically allocated, the problem of unbalanced information utilization between modalities is solved, and the accuracy of sentiment analysis is improved.
Patent Information
- Application Number
- CN202510818757.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-18
- Publication Date
- 2025-09-12
AI Technical Summary
In existing multimodal sentiment analysis methods, the feature differences between modalities are not fully utilized, resulting in unstable sentiment analysis results. The traditional equal-weighted fusion method cannot effectively combine the contributions of different modalities.
A multimodal sentiment analysis method with cross-modal supplementation and balance is adopted. By constructing a feature extraction module, a dominant modality supplementation module, a cross-modal comparative analysis module and a balanced prediction result module, cross-modal comparative learning is used to enhance modal relevance and difference, and dynamically allocate the importance of prediction results.
It improves the accuracy of sentiment analysis, solves the problem of unbalanced information utilization between modalities, and achieves the stability of multimodal collaborative enhancement and feature fusion.