Research On Weibo Rumum Test Method Based On Deep Migration Learning

The spread of rumors in Weibo has severely affected people's daily life and the healthy development of social networking platforms.But there is only a small amount of labeled data in the field of Weibo rumor detection.Aiming at the problem of insufficient labeled data,two deep transfer learning methods are proposed,and two models are designed and implemented:(1)Weibo rumor detection method based on model migrationA joint model(Transferring Learn-BiGRU-2-CNN,TB2GC)composed of a two-layer bidirectional gated recurrent unit(BiGRU)and a convolutional neural network(Convolution Neural Network,CNN)is designed as a feature extractor.First,use the rich comment data in other fields to pre-train the joint neural network,and then the feature extraction layer in the trained model is migrated to the Weibo rumor detection task.In order to adapt to the target task and fast convergence,the feature extraction layer is adjusted by distinguishing fine-tuning and oblique triangle learning rate,and the feasibility and effectiveness of the method are proved by experiments.(2)Weibo rumor detection method based on multi-task learningTaking into account the opposite emotional tendencies between rumors and non-rumors,a multi-task learning framework(BERT-BiGRU-MTL,BBi GM)combined by BERT and BiGRU is designed.In joint training,the common features between tasks and specific features for rumor detection tasks are extracted at the same time,sentiment analysis tasks are used to assist rumor detection tasks,and the feasibility and effectiveness of the method are proved through experiments.