Group 4 (2025 MSwAI) - Changes

[UnleqitQ]

This pull request introduces support for new annotation categories—Offensive Speech, Emotions, Toxic, and Sentiment—across the corpus annotation UI, document reader, and lexicon. It also adds sidebar model selection controls and associated styling, and updates the document reader to handle these new categories both visually and functionally.

The most important changes are:

Support for New Annotation Categories:

• Added UI elements in corpusAnnotations.ftl to display the presence of Offensive Speech, Emotions, Toxic, and Sentiment annotations as checkboxes.
• Updated lexicon.css to style lexicon entries for the new annotation types: sentiment, emotion, offensivespeech, and toxic. [1] [2]

Document Reader Enhancements:

• Introduced new CSS classes in document-reader.css to visually distinguish text spans and markers for emotion, offensive, toxic, and sentiment annotations within the document content. [1] [2]
• Added JavaScript logic to support sidebar model selection, including retrieving selected models, triggering page reloads on model changes, and marking visualization panels as dirty when emotion models change.
• Modified the lazy loading of document pages to include the selected model configuration in API requests, ensuring the displayed content matches the user's model selection. [1] [2]

Sidebar and Visualization Updates:

• Added sidebar model selection UI and corresponding CSS for a more user-friendly experience when choosing models for each annotation category.
• Extended visualization panel and navigation logic to support a new panel (vp-6) for emotion development visualization. [1] [2]

General UI Improvements:

• Added a .flexed-column utility class for flexible column layouts.