Skip to content

Home

Jump to bottom
Julian Wendlandt edited this page Jul 7, 2026 · 4 revisions

Welcome to the Automated Thematic Analysis Wiki! 🚀

Welcome to the official wiki for the Automated Thematic Analysis project (AMOS SS 2026).

This project aims to run qualitative data analysis by automating the extraction and thematic categorization of knowledge from unstructured text, such as interview transcripts. By using Large Language Models (LLMs), our system maps text chunks to a hierarchical thematic tree, building a robust thematic knowledge graph.

🌟 Key Features

  • Traceable Pipeline: Extract quote-grounded codes from interview transcripts and synthesize them into a reviewable codebook — every theme/code traces back to an exact transcript quote.
  • Codebook Generation & Application: Generate a new codebook from your transcripts with an LLM, or deductively apply an existing codebook to new transcripts as a repeatable, exportable analysis run.
  • Thematic Knowledge Graph: Explore theme/code frequency, interview coverage, and demographic breakdowns through an interactive theme browser.
  • Advanced AI Integration: Utilize LangChain to support models (e.g., GPT-OSS, Gemma) via the FAU NHR gateway or GWDG Academic Cloud, switchable at runtime from the UI.
  • Modern Architecture: Built on top of FastAPI, PostgreSQL (async), and Pydantic, fully containerized with Docker for easy deployment.

📚 Quick Links

These pages will help you get started:

  • Build & Deploy: Learn how to quickly spin up the project locally using our bootstrap scripts and Docker.
  • User Documentation: A walkthrough of the core features and the end-to-end workflow.
  • Backend Routes: Overview of backend API route groups and links to detailed route documentation.
  • System Architecture: Dive into the system's core components, including the layered architecture and persistent data model.
  • Frontend and Routing: The Flask UI's blueprints, routes, and design system.
  • LLM & Infrastructure: Understand our model selection strategies, LangChain integration, and how to run analysis on environments like the FAU GPU Cluster versus the Academic Cloud.

Feel free to navigate through the sidebar to find more detailed documentation on specific components of the platform.

Clone this wiki locally