Bert Attention-based BiLSTM Intelligent Mode for Fake Text Detection

Bert, BiLSTM, and Attention mechanisms are key techniques in the field of machine learning. Currently, BERT and BiLSTM models are commonly applied to multiple tasks in the field of Natural Language Processing (NLP), including text classification, named entity recognition, relation extraction, question-answer systems and so on. Additionally, Attention mechanisms have also been widely used in the fields of image, NLP, and reinforcement learning to enhance the expressive power of the models. Therefore, this paper proposes a Bert Attention-based BiLSTM intelligent model for fake text detection and constructs the dataset Detection Dataset as a test dataset to compare the effectiveness for fake text detection of four models. These four models are Bert-LSTM, Bert- BiLSTM, Bert Attention-based LSTM and Bert Attention-based BiLSTM model. The experimental results show that the accuracy of the Bert Attention-based BiLSTM model proposed in this paper is 90.24%, which is significantly higher than the other three models for text detection accuracy. This model uses the Bert pre-trained language model to represent the text in a vectorized manner to construct a transfer model, combines the attention mechanism to dynamically assign weights to the text, and finally uses the BiLSTM to perform binary classification to achieve efficient judgment of web text authenticity.