Annotation Guide

Learn how to annotate text with frame semantic information.

Annotation Types

Webtool 4.2 supports multiple annotation modes:

Full-Text Annotation

Annotate complete sentences with frame and frame element information.

1. Select a sentence from your corpus
2. Identify the target word (lexical unit)
3. Mark frame elements in the sentence
4. Assign appropriate FE labels

Multimodal Annotation

Annotate images and videos with semantic information:

• Static Objects: Bounding boxes on images
• Dynamic Objects: Time-based regions in videos
• Deixis Annotation: Spatial and temporal references

Annotation Workflow

1. Select Corpus → 2. Choose Document → 3. Pick Sentence
                     ↓
4. Identify Target LU → 5. Mark Frame Elements → 6. Save Annotation

Quality Guidelines

• Ensure frame elements don't overlap
• Use the most specific FE labels available
• Annotate all core frame elements when present
• Mark null instantiations explicitly
• Review annotations for consistency

Keyboard Shortcuts

• Ctrl+S: Save annotation
• Ctrl+Z: Undo last change
• Ctrl+Y: Redo
• Esc: Cancel current annotation

Annotation Reports

Generate reports to track annotation progress:

• Annotations by frame
• Coverage statistics
• Inter-annotator agreement
• Quality metrics

See Also

• Frame Elements
• Corpus Management