Research on false information detection algorithm in external publicity based on interactive relationship context

The innovative integration of computer technology and interactive relationship context technology has opened up a new path for the detection of false information in external publicity. To meet the needs of quickly distinguishing false information from real information and accurately detecting it in different scenarios, this study proposes an information detection model combining Generative Adversarial Networks (GAN) and Graph Attention Networks (GAT). The model implements adversarial training and learning weight allocation to achieve accurate identification and detection of false information. Test results show that the model's accuracy rate for identifying different types of information is over 98%. In the training set test, the precision rate of the model is 99.56%, the recall rate is 99.43%, and the F1 score is 0.9891. Experimental results indicate that the proposed model can accurately capture false information in external publicity, effectively improve the accuracy of false information identification and detection efficiency, and provide a new approach for the application of artificial intelligence technology in the detection and analysis of false information in external publicity.