Table 1.
The different models and datasets based on pattern-based matching.
Table 2.
The different models and datasets based on machine learning.
Table 3.
The different models and datasets based on deep learning.
Fig 1.
The overall model framework.
Fig 2.
The semantic information in the sequence.
This example come from SemEval2010 Task8.
Table 4.
The experimental environment and illustration.
Table 5.
The hyper-parameters of Bert for fine tune in SemEval-2010 Task-8 dataset.
Fig 3.
Precision, recall and F1 rate of piecewise convolution and pooling module.
Fig 4.
Precision, recall and F1 rate of focal loss module.
Table 6.
The comparison on test results of different modules.
Fig 5.
Comparison on precision, recall, F1 results of different modules.
Table 7.
Comparison of R-BERT and proposed model.
Fig 6.
The influence of this module on the precision, recall and F1 of every class.
Table 8.
The hyper-parameters of Bert for fine tune in SemEval-2018 Task-8 dataset.
Fig 7.
Comparison on precision, recall, F1 results of different modules.
Fig 8.
F1 variation diagram of different model on two datasets.
Table 9.
The comparison of results of different models.