An Entity Relation Extraction Method for Few-Shot Learning on the Food Health and Safety Domain

Table 4

Detailed parameters of the experiment.

Model	Construction and output	Training hyperparameters
Text noise removal classifier	Input; BERT character; Bi-LSTM layer  8; CRF	Batch size: 128; Learning rate: 0.001; Epochs: 100; Loss function: negative log-likelihood; Update strategy: Adam
Feature enhancement	Input; Convert function; BERT character	—
Chars and words fusion	Input; BERT words; Combine function sigmoid × 2	Batch size: 128; Learning rate: 0.001; Epochs: 200; Loss function: binary cross-entropy