Chinese Shallow Semantic Parsing Based On Multi-Method Of Machine Learning

With the rapid development of 5G+ information intelligence, higher requirements are put forward for accurate and efficient semantic annotation methods. Semantic role annotation for any single method at present has its obvious and complementary advantages and disadvantages. Therefore, this paper attempts to introduce the above three mainstream and stable annotation methods into each task of semantic role annotation, and designs a Chinese semantic role annotation that integrates multi-method. This method integrates the statistical-based linear sequence method, the rule-based hierarchical tree method and the most advanced deep learning in the four processing modules of semantic role annotation. Multi-level linguistic features are introduced into the feature arrangement of the model to realize the mutual combination of multiple modules. Experiments show that the modular fusion of steps and methods effectively improves the annotation performance of each step of annotation.