| With the popularization of the Internet,especially the rapid development of mobile Internet in recent years.More and more people express their opinions and opinions through the Internet,thus producing a large number of text messages with emotional orientation.These text messages are of great practical value for business decisions,public opinion control and other applications.How to excavate emotional information in texts has gradually become the research direction of a large number of researchers.Text emotion classification includes many research fields,among which,the task of fine-grained text emotion classification is the current research hotspot,and fine-grained text emotion classification is to judge the emotional tendency of text at a given Angle.This paper analyzes the advantages and disadvantages of the existing fine-grained text emotion classification methods based on the study of a large number of relevant literature at home and abroad.Based on the previous research results,this paper proposes a fine-grained text emotion classification method using linguistic constraints and attention mechanism.The method mainly includes three steps: first is the text by LSTM length memory network modeling,the semantics of the text,according to the following for a given Angle is obtained by using the attention mechanism of semantic weight,get a weight of text semantic representation,finally,the use of linguistic resources to construct the four linguistic constraints,and combing linguistic constraints and attention mechanism is introduced into the model of the loss function,participate in the training model,further optimization model of the effect.Proposed in this paper the text of the fine-grained emotion classification method is different from previous studies,this article first to linguistic constraints and attention mechanism together,experimental results showed that our model can be captured in the emotional expression with the given Angle related language character,have certain theoretical significance and practical application value,for the next research provides a train of thought enlightenment. |