cj-dev-code/semantic_search

1---2title: Semantic Book Searcher3emoji: 📚4colorFrom: gray5colorTo: blue6sdk: docker7app_file: Dockerfile8pinned: false9---10 11 12 13# 🧠 Mental Health Semantic Searcher14 15A project to make mental health advice from dense books more accessible through semantic search and quote-level retrieval.16 17## ✨ Motivation18 19When you're dysregulated, skimming a whole book or even a chapter can feel overwhelming. This tool helps you search for **therapeutically helpful language** based on what you're feeling or trying to express — so you can find quotes, not just concepts.20 21---22 23## ⚙️ Tech Stack24 25- **Python** – backend logic + semantic chunking26- **Azure AI Search** – vector search. rejected: Pinecone, FAISS, Vilvus27- **Vector Embeddings** - voyage 3.5. chosen for: intent, emotional, concept rich embedding. rejected: openai (too literal), cohere (unemotional), MPnet MiniLM (poor purpose handling), Google USE (too generic for quote level lookup)28- **Gradio** – frontend interface. Options rejected: streamlit, dash, next.js, react.29- **GitHub Actions** – CI pipeline30- **pytest** – test framework31 32---33 34## 📂 Project Structure35 36```37frontend/    # User interface (Streamlit or React)38backend/     # API + semantic search logic39data/        # Sample books and chunked quote data40notebooks/   # R&D, experiments, visualizations41scripts/     # Setup scripts, evaluators42```43 44---45 46## 🚧 Status47 48- [x] Repo initialized with CI49- [x] Project scaffolded50- [ ] Semantic chunker implemented51- [ ] Retrieval working end-to-end52- [ ] Frontend MVP53 54---55 56## 🧪 Running Tests57 58```bash59pip install -r requirements.txt60pytest61 62---63 64## 🔐 Environment Variables65 66Before running the app, create a `.env` file in the project root and define the following:67 68| Variable           | Description                          |69|--------------------|--------------------------------------|70| `VOYAGE_API_KEY`   | API key from [Voyage AI](https://voyageai.com) |71| `AAIS_ENDPOINT`    | Azure AI Search endpoint URL         |72| `AAIS_KEY`         | Azure AI Search admin key            |73| `OPENAI API KEY`   | Openai API key                       |74| `RUN_DOCKER_TEST`  | make this true to run docker tests   |75 76You can use the provided `.env.example` file as a template:77