Text Sentiment Analysis For Education Public Opinion

Education emphasizes both "national economy" and "people’s livelihood." In recent years,the internet has provided a more convenient channel for the people to gather public opinion and express their demands,and the public’s attention to education has continued to increase.The interaction between online education public opinion and real social education has a profound impact on real education.Making full use of online education public opinion is conducive to education managers to make scientific decisions and enhance the public’s ability to monitor education.Emotional orientation is the vane of online education public opinion,and it is a true reflection of the opinions and attitudes of all levels of society.The data generated is mainly text,which contains short text comment data and longer news texts,which are important value.This emotional information affects the trend of online education public opinion.In order to prevent and control the occurrence of negative public opinion and maintain the healthy operation of national education,how to accurately mine the emotional semantic information in online education public opinion text is a research difficulty in the field of sentiment analysis.Text sentiment analysis based on deep learning is a popular and effective method,which can automatically perform sentiment feature generation stage to mine text sentiment feature information.In this paper,two different texts,short text comments and long news texts generated in online education public opinion,are used as the practical application background,combined with deep learning technology,to carry out research on sentiment analysis methods for texts of different lengths.Existing short text sentiment analysis methods cannot extract text sentiment features more comprehensively,and rely heavily on a large amount of language knowledge and sentiment resources.It is necessary to make full use of these unique sentiment information to achieve the best performance of the model.This paper proposes a text sentiment analysis capsule model that combines convolutional neural networks and Bi-GRU networks.The model first uses multi-head attention to learn the dependencies between words,captures emotional words in short texts,uses convolutional neural networks and Bi-GRU to extract emotional features of different granularities in the text.After the feature fusion,the global average pooling is used to get the instance feature representation of the text;meanwhile,attention mechanism is combined to generate feature vectors for each emotion category to construct an emotion capsule.Finally,the emotion category of text is judged by the capsule attributes.Experimental results show that this model is better than other baseline models on short text sentiment classification tasks.In the sentiment classification task of long texts,feature extraction is relatively difficult due to the long text.Existing long text sentiment analysis methods do not take into account the important role of text hierarchical structure in sentiment classification,and cannot effectively encode and learn the spatial relationship between the part and the whole of the text.In response to these problems,this paper proposes a text sentiment classification model based on a hierarchical self-attention mechanism capsule network.The model first inputs word vectors into a hierarchical self-attention network based on a recurrent neural network neural network,extracts broader semantic dependencies from two levels of words and sentences,and pays attention to important words and sentences in the text.The result is input into the capsule network to encode the relationship between the part and the whole,so as to make full use of the feature information of the whole text,which is beneficial to improve the accuracy of sentiment classification tasks.In addition,the attention mechanism in traditional hierarchical networks requires more parameter dependence and cannot make the model better capture the dependence between words,this paper proposes that hierarchical self-attention networks can reduce external parameter dependence and encode text semantics.Experiments prove that this mechanism can effectively improve the classification performance.Experimental results show that the model can learn the logical relationship between contexts from long texts,and effectively find the key information with the greatest emotional semantic contribution.Through the comparison with the baseline model,the effectiveness of the model on the task of sentiment classification of long texts is verified.