Figure6

Figure 6. Training results. (a) is the F1 score curve, with the horizontal axis representing the number of iterations of SSL and the vertical axis representing the F1 Score of the model for identifying named entities. During all iteration cycles, there is only one input instruction for the model. (b) is the sampling of the NER results of the model in the training process, which is used to record the named entity sequences generated by SCNER under different training cycles. In the figure, the label of a named entity with green background is identified correctly, whereas the label without processing is identified wrongly, and its display content is consistent with the F1 Score curve.