Text Sentiment Analysis And Research For Social Media Reviews

In recent years,with the rise of social media,a large number of users have been active on various social media platforms every day,such as Xiaohongshu,forums,Weibo,and shopping websites with social functions,generating a large number of comments on things,products,or services that users participate in.These comments often contain rich sentiment information.If we can mine users’ sentiment tendencies based on these comment texts,it will play an important role for both the platform and users.However,due to the strong specificity of existing text comment data,short text length,and multiple specific aspects of sentiment polarity,traditional coarse-grained sentiment analysis can no longer meet people’s needs.Aspect level sentiment analysis,as a fine-grained sentiment analysis task,can deeply explore the hidden associations between aspect words and context,thereby determining the sentiment polarity of different aspect words in a sentence.Based on this,this article takes social media comment data as the research object and conducts research on aspect level sentiment analysis.The main research work is as follows:(1)Most aspect level sentiment classification models based on recurrent neural networks and attention mechanisms rarely consider the syntactic structure information between context and aspect words,and ignore the interaction information between context and aspect words.In response to the above issues,this article proposes an aspect level sentiment classification model based on graph convolutional networks and interactive attention.The model first uses Bi LSTM to extract the hidden features of the context,then uses graph convolution neural network to extract the syntactic structure information in the sentence according to the syntactic dependency tree,then uses the interactive attention mechanism to learn the interactive information between the aspect word and the context,and finally performs feature fusion,output and classification.The effectiveness of this model was verified through experimental comparisons with other models in a series of public datasets published by Sem Eval.(2)At present,graph convolutional network models based on dependency trees focus on grammar information when modeling,lacking in mining sentiment knowledge in sentences.In response to this issue,this article proposes an sentimen enhanced dual channel graph convolutional aspect level sentiment classification model.The model conducts research from both semantic and grammatical perspectives.Firstly,Bi LSTM is used to encode the sentence word vector and obtain the hidden layer representation of the sentence.Then,in terms of syntax,the sentiment knowledge from Sentic Net is fused on the syntactic dependency tree to enhance the dependency relationship of the sentence,so that the model considers the grammar information of contextual words and aspect words,as well as the mining of sentiment knowledge in the sentence;Semantically,by using the attention mechanism to learn the semantic correlation between words and the context in the sentence,the model can pay attention to the Semantic information in the sentence while paying attention to grammar.Finally,perform interactive learning between the two modules and output classification.The effectiveness of this model was verified by comparing it with other model experiments on publicly available datasets.