A Study of Using Principal Component Analysis to Enhance Semantic Role Annotation in Chinese Autoanalysis

The article explores the ways to improve the effect of semantic role labeling in the process of automatic Chinese language analysis.The traditional calculation methods for information content, distance and attributes are improved, and the dynamic weight calculation method for semantics is proposed and comprehensively weighted based on principal component analysis.A Chinese semantic role labeling model based on Highway Bi-LSTM is designed, and Self-Attention is added to improve the information representation of semantic role labeling.The correlation coefficient of this paper's method with the manual judgment value is above 0.9, and the average Kappa coefficient is 82.43%, which is very close to the effect of manual annotation.The F1 values of this paper's model are improved by 18.56% and 8.95% respectively after adopting the attention mechanism, indicating that the design of this paper's method is reasonable.It can be seen that the method of this paper helps to improve the effect of semantic role labeling.