Research on intelligent identification algorithm of digital platform practitioners based on multidimensional feature fusion

Traditional identification methods mostly rely on a single feature, which is difficult to cope with the cross-platform flow and dynamic changes of practitioners' identities, resulting in low identification accuracy and weak generalization ability. To solve these problems, this paper proposes an intelligent identification algorithm for digital platform practitioners based on multi-dimensional feature fusion. The algorithm constructs a multi-dimensional feature space including static attribute features, behavior sequence features, cross-platform correlation features and biological features, and conducts cross-modal feature alignment through a comparative learning method to optimize the semantic consistency of the feature space. At the same time, a dynamic weight allocation mechanism is designed, which combines the context-aware attention model and space-time factors to correct the weight and improve the contribution of key features. In addition, the graph convolution network (GCN) is introduced to enhance feature fusion. By constructing a cross-platform association graph, high-order association patterns between nodes are mined to identify potential false identity clusters. The experimental results show that the proposed method is superior to the traditional baseline model and the deep learning contrast model in all indicators. The accuracy on the test set is 92.3%, F1-score is 90.5%, AUC-ROC is as high as 0.941, and the recall rate of cross-platform forged identity is 89.4%, which is significantly superior to other models, which verifies the effectiveness and robustness of this method in identity recognition tasks.