Digital-Liahona

A mobile scripture study companion that surfaces spiritually relevant passages using a Neo4j knowledge graph, semantic embeddings, and a lightweight LLM backend.

---

Table of contents

1. Features
2. Architecture
3. API Reference
4. Acknowledgements

---

Features

Category	Highlights
Natural‑language Q&A	Ask "How do I find peace?" and receive scripture verses and General Conference insights.
Hybrid retrieval	Vector similarity + topic + keyword search for high recall, then LLM re‑ranking for precision.
Knowledge graph	Verses, talks, topics, speakers, cross‑references stored as first‑class nodes in Neo4j.
Local‑first ML	Embeddings via nomic‑embed‑text; answer ranking with llama3.2 (runs in Ollama).
React Native app	Cross‑platform chat UI; tap a verse to open the full chapter; dark‑mode & bookmarking on roadmap.

---

Architecture

┌────────────┐   POST /chat        ┌───────────────┐
│ React Native│ ───────────────▶ │  Flask API    │
│   (iOS/Android) │◀───────────────│  (Python 3.12.3) │
└────────────┘   JSON reply       │  ├─ Vector search in Neo4j
                                   │  ├─ Topic/keyword match
                                   │  └─ LLM re‑rank (Ollama)
                                   └───────────────┘
                                            │
                                  ┌─────────────────┐
                                  │  Neo4j 5 graph  │
                                  │ Verse ↔ Topic ↔ Talk │
                                  └─────────────────┘

Backend endpoints (excerpt):

• POST – main Q&A route
• GET – read‑only chapter view
• GET – verses & talks for a gospel topic
• GET – full talk transcript

Data is populated via process_import_data_neo4j.py, which:

1. Cleans scripture & talk JSON
2. Creates graph schema + vector indexes
3. Embeds verses & paragraphs

---

API Reference

POST /chat

// Request
{ "message": "How does Christ teach forgiveness?" }
// Response (truncated)
{
  "reply": "Here are some things that might help:\n...",
  "verses_with_links": [
    {
      "text": "...",
      "source": "Matthew 18:21",
      "chapter_info": {
        "chapter_url": "/chapter/matthew/18"
      }
    }
  ],
  "related_content": { ... }
}

Refer to flask_server.py for additional endpoints.

---

Acknowledgments

• Open Scriptures API for scripture data
• The Church of Jesus Christ of Latter‑day Saints (General Conference transcripts)
• Neo4j & Ollama communities for open tooling