In view of the low accuracy of Internet rumor detection in traditional machine learning detection methods due to the influence of comment directivity and incomplete features caused by too long sequences, an Internet rumor detection method combining attention mechanism and bidirectional gated recurrent unit (GRU) model in big data environment is proposed. First, the continuous bag of words (CBOW) model is modified by defining the context window and introducing the weighting module, and it is used to obtain the word vectors of the text. Then, the word vectors are used as the word embedding layer which is fed into a bidirectional GRU combined with an attention mechanism through the convolutional layer and pooling layer of the convolutional neural network (CNN) to fully extract the features of the text. Finally, the text feature vectors are input into the Softmax classifier for classification so as to realize the detection of Internet rumors. Based on the selected microblog dataset, the experimental results of the proposed model show that when the learning rate is set to 0.015, Accuracy, Precision, Recall, and F1 values are about 95.8, 93.1, 92.9 and 93.5, respectively, which are better than other comparison models.
In order to explore whether pre-embedding can better support learning in VR videos, this paper takes 30 college students as the research object, and conducts a single-factor randomized experiment of “pre-embedding problems and non-embedding problems in VR videos” The results show that pre-problems can effectively increase the time learners pay attention to the corresponding learning content, have a positive impact on learning retention and long-term memory, and increase the cognitive load in the learning process and have no significant impact on knowledge transfer. The research conclusions of this article can provide reference for the development of VR video teaching resources.