MultiMedia Annotator (MM-Annotator)¶
Web frontend for the TextAnnotator backend — a browser-based platform for collaborative, real-time annotation of text and multimedia (image) documents, with optional LLM-assisted annotation.
Built with React 19, TypeScript, Vite, Tailwind CSS v4 and shadcn/ui. State is held in Zustand; all live annotation traffic runs over a single WebSocket to the TextAnnotator UIMA CAS service.
Features¶
- Project workflow — browse projects, track per-document status and overall progress, resume annotation. See Annotation.
- Real-time collaboration — persistent WebSocket, batched CAS edits, live updates. See API & WebSockets.
- Criteria-based rating forms — configurable button/select/numeric sections
mapped to UIMA
Categoryannotations. - Multimedia — text, Q/A scenarios and virtualised, lazy-loaded image galleries side by side.
- LLM assistance (UCE RAG Bot) — streaming chat with text/vision models that can auto-fill rating fields. See RAG Bot.
- Admin tooling — upload, validation, permissions, export. See Admin.
- Role-based access and dark mode.
Quick Start¶
See Installation and Configuration for full setup.
Demo¶
A live demo is available at eval.textannotator.texttechnologylab.org.
| Username | DemoAnnotator |
| Password | demo2025 |