一种基于改进的动态图神经网络的不实信息检测方法
By improving the dynamic graph neural network and combining the encoder, decoder and attention mechanism, the problem of ignoring the temporal dynamics and temporal sequence information relationship in misinformation detection is solved, and more efficient and accurate misinformation detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING UNIV OF POSTS & TELECOMM
- Filing Date
- 2023-11-13
- Publication Date
- 2026-07-17
AI Technical Summary
Existing methods for detecting misinformation fail to effectively consider the dynamic nature of misinformation dissemination over time and ignore the relationship between the dissemination structure and temporal information, resulting in poor detection performance.
An improved dynamic graph neural network is adopted. By acquiring user-posted post data, it is divided into a time-series graph of false information data of length T. Spatial features are extracted and reconstructed using an encoder and decoder. Combined with the dynamic graph neural network model, an attention mechanism is used for feature weighting. The model is trained using cross-entropy loss, reconstruction loss and KL divergence loss.
This improves the accuracy and efficiency of the model in detecting misinformation, enabling it to better capture the correlation between time dependencies and node features, construct more accurate time-series information, and enhance the effectiveness of misinformation detection.
Smart Images
Figure CN117392686B_ABST