News Content Judgment Research Based On LSTM Algorithm

Generally speaking,fake news is a false narrative that is published and promoted as something real.Social media now creates an environment for fake news,where people can post fake news for profit,or automatically generated fake news through bots.News consumption on social media is a double-edged sword.On the one hand,it consumes news with low cost,easy access and fast dissemination;on the other hand,it enables the widespread dissemination of fake news,i.e.,low-quality news that deliberately provides false information.The media plays an important role in the public dissemination of information about events,and the rapid development of the Internet allows information to spread quickly through social networks or websites,where unverified or false news spreads through social networks and reaches thousands of users without concern for the credibility of the information.Fake news is usually generated for commercial and political gain to mislead and attract readers.The widespread dissemination of fake news has the potential to have an extremely negative impact on individuals and society.As a result,fake news detection on social media has recently become an emerging research that has attracted tremendous attention.Fake news detection on social media presents unique features and challenges that make existing detection algorithms in traditional news media ineffective or inapplicable.First,fake news is intentionally written to mislead readers into believing false information,which makes news content-based detection difficult and uncomplicated;therefore,we need to include auxiliary information,such as users’ social engagement on social media,to help make judgments.Second,leveraging this ancillary information is inherently challenging because the data generated by users’ social engagement with fake news is huge,incomplete,unstructured,and noisy.Automatic trustworthiness analysis of news articles is a current research hotspot,and deep learning models are widely used for linguistic modeling.Typical deep learning models,such as convolutional neural networks(CNN)and recurrent neural networks(RNN)can detect complex patterns in text data.Long Short Term Memory(LSTM)is a tree-structured recurrent neural network for analyzing continuous data of variable length.The bi-directional LSTM allows to view specific sequences from front to back as well as from back to front.In this thesis we use python as the development language and also use the sklearn toolkit to build out random forest,logistic regression and LSTM models,complete fitting and training of the models,compare and analyze the results of different models using the tensorflow toolkit,and propose a Bidirectional-LSTM recurrent neural network based on fake news The reliability of the detection model.In this thesis two publicly available unstructured news article datasets from Kaggle are used to evaluate the performance of the model.The findings of this thesis are that the results of the model detection show that the LSTM model is more accurate than other methods,i.e.Random Forest,Logistic Regression,for fake news detection.