Research On Sentiment Classification Using User Information

With the rapid development of social media,it is common for users to express their feelings and opinions on various social platforms.How to effectively utilize the data of these platforms for sentiment analysis has been a hot topic in natural language processing research.Among them,using comment texts to perform sentiment classification is one of the main tasks in sentiment analysis.At present,most of the current research on sentiment classification is to identify sentiments based on the text's information.However,on social media,user information is very helpful for identifying emotions.Thus,this paper makes in-depth research on how to extract the relationship between users,how to construct an effective sentiment classification model,and how to integrate user information into sentiment classification.The specific research content is divided into the following three aspects:First,to make better use of user information and build a network of associations between users.This paper proposes a neural network model to identify the friend relationship between users from the text of user reviews.This article first combines user review text and user attribute information on social media to build user feature representations.Second,it integrates user feature representations into neural network models,using convolutional neural networks and attention mechanisms to analyze two users' relationship.Experiments show that the model proposed in this paper can improve the performance of user friend relationship recognition.Secondly,to better build a sentiment classification model based on user information,this paper proposes a sentiment classification model based on text hierarchy.This paper constructs a hierarchical structure of user review texts and learns such a structure based on a neural network model,thereby establishing a sentiment classification model based on a hierarchical neural network.Experiments show that the model proposed in this paper can effectively improve the performance of sentiment classification and prove the effectiveness of the text hierarchy for semantic modelling of user reviews.Finally,to integrate user information into the product emotion classification model,this paper proposes a product emotion classification model based on multiple outputs and hierarchies.Specifically,first,a review representation model based on text and user information is constructed.Second,while classifying the overall product sentiment tendency,the user's review text sentiment tendency is also predicted,so that the influence of user information on sentiment classification is considered in both the presentation layer and the output layer.Experiments show that the model proposed in this paper can effectively improve the performance of sentiment classification of the overall product,and prove that user information is of great help to product sentiment classification.