Nested named entity recognition in traditional Chinese medicine electronic medical records via dual-granularity feature augmentation and span classification.

As can be seen from the figure, all the equivalent models have lower F1 index values than DG-SpanTCM in most of the cases.As the training process continues, the values of DG-SpanTCM stabilize and show an overall increasing trend without overfitting, while this is not the case for some of the comparison models, which show a decreasing trend, i.e., there is an overfitting problem, after the model stabilizes for a period of time.

also in 47,180 other papers