Step 1: Creating a new project
Step 1: Creating a new project
- Go to
Admin Settings -> Projectsand clickCreate new project - Enter the project name and description and click
Create - Click on the newly created project to go to the project dashboard
Step 2: Upload data
Step 2: Upload data
- Go to
Uploadsin the project dashboard, clickCreate an upload. - Upload native documents, see import-native-documents.md.
- Upload preprocessed documents, see import-preprocessed-documents.md.
- Upload annotations, see import-annotations.md.
Step 3: Upload a review protocol
Step 3: Upload a review protocol
- Go to
Protocols -> Review Protocols, clickCreate new protocol. - Upload an existing review protocol or create a new protocol from a template. See review-protocols.md.
- Set the main protocol that contains instructions for review as the
defaultprotocol.
Step 4: Model training/predictions
Step 4: Model training/predictions
- Go to
Modelsand click on a pre-trained model of interest. - Clicking the
Predictionstab to view model predictions. See classification-models.md. - Training a model by updating the model instructions or adding annotations. See classification-models.md.
Step 5: Create review batches
Step 5: Create review batches
- Go to
Review -> Review Batchesand clickCreate review batch. - Create an active learning batch for model training or a control set batch for model evaluation. See review-batches.md.
- Go to
Review -> Review Assignmentsand click the corresponding review batch to start annotating.
Step 6: Monitor performance
Step 6: Monitor performance
- Go to
Model -> Predictionsand clickMetrics. - Selecting a classification threshold based on the Precision and Recall tradeoff.
- Click
Update predicted tagsto update model predictions using the selected threshold.
Step 7: Export results
Step 7: Export results
- Go to
Exportsand clickCreate an export. - Export annotation results. See exports.md.
- Export documents. See exports.