Research On Network Text Sentiment Classification Based On Deep Learning

With the popularization and development of the network,more and more users use Internet to exchange information and publish opinions.A large amount of text information is flooded in the network environment because of the openness and inclusiveness of network.Timely monitoring and analysis of the emotional trend of network information has a great significance to the security of network public opinion.But it is almost impossible to classify and identify the information and opinions by human beings,because it will take a lot of time and energy.However,if we want to make use of the common classification technology such as keyword classification,the accuracy of classification will be not accurate enough due to the complexity of language expression.This article combines the important research directions of current computer fields,security and intelligence,aiming at dealing with the content security problem of massive data,using the related technologies of Natural Language Processing,training and learning network text samples by adopted the deep learning model to achieved the efficient processing of massive network text sentiment information.Long Short-Term Memory(LSTM)has obtained strong results on a variety of sequence modeling tasks because of its superior ability to preserve sequence information over time.Focus on the diversified opinion expression form and the explosive growth of information amount in network environment,we propose a text emotion recognition model based on multi-dimensional LSTM in order to improve network content security by making full use of additional information of text samples.In this paper,we divide the original sample into two parts: the main information sample and the additional information sample.Then multi-dimensional LSTM model is used to extract the features vectors of them.Finally,according to the results of the two feature vectors,the sentiment classification result is carried out by feature fusion and further computation.The multi-dimensional LSTM model is implemented and tested by TensorFlow.And the experimental results show that the emotion recognition classification accuracy has been greatly improved by taking advantage of multi-dimensional LSTM.