Environmental False Complaint Detection Based on Adversarial Transfer Learning Model
-
Graphical Abstract
-
Abstract
To achieve environmental false complaint report detection, a false complaint and report detection model was proposed based on adversarial transfer learning method. First, a long-short term memory (LSTM) network was used as a feature extractor to extract the shared features of Weibo rumors (source domain) and complaint report text (target domain). Second, the domain adaptation was performed by using the adversarial learning method to align the source domain features with the target domain features. Finally, the classification results were output by the classifier, and the network parameters were updated by the classification loss and the domain adaptation loss. Model comparison experiments and ablation experiments were designed, and the F1 value of the model reached 79.61%, indicating that the adversarial transfer learning model has good performance and is suitable for application in the the task of detecting environmental false complaints and reports.
-
-