After you build the dataset, you can start training the model. Return to your project, switch to the Model Hub tab, and click Create Model.
On the Create Model page, the platform handles the implementation details. To begin, enter a name for the model.
For Model Type, you can choose from several model types that support Chinese or English entity extraction. For more information, see Model descriptions.
The system selects the Chinese LSTM-CRF model by default. Click the Add Training Data button to select the datasets that you have annotated or uploaded. Then, click Submit. The model automatically starts training. The Create Model page also provides other options, such as uploading a custom dictionary, setting regular expressions, and advanced settings. These options are described in detail in the "Rules Engine and Advanced Settings" chapter.
After the model finishes initialization, it automatically enters the training phase. Click View to open the model details page, where you can view training logs and specific evaluation metrics.
The evaluation metrics for entity extraction are Precision, Recall, and F1-score. The value for each of these metrics ranges from 0 to 1. A higher value indicates better model performance.
Precision: For a single class, this is the ratio of correctly predicted samples to the total predicted samples for that class. For the entire model, it is the ratio of all correct predictions to the total number of predictions.
Recall: For a single class, this is the ratio of correctly predicted samples to the total number of actual samples for that class. For the entire model, it is the ratio of all correct predictions to the total number of samples.
F1-score: The harmonic mean of precision and recall.