SLR Platform — Walkthrough

What Was Built

A production-grade AI-powered Systematic Literature Review platform with:

Backend (FastAPI + SQLAlchemy)

Layer	Files	Description
Models	8	User, Paper, Chunk, ExtractionSchema, ExtractionResult, Matrix, AuditLog, ZoteroMapping
Routers	7	Papers, Zotero, Extraction, Matrix, Analytics, Clusters, Conflicts
Services	12	PDF parser, chunker, section classifier, extraction engine, confidence scorer, LLM client, Zotero sync, SLR protocol, matrix builder, analytics, clustering, conflict detection
Schemas	4	Pydantic models for all API request/response types
Utils	1	Provenance tracking and source highlighting

Frontend (Next.js + Tailwind)

Page	Route	Features
Dashboard	/	Stat cards, quick actions
Papers	/papers	Upload, filter, process, delete
Extraction	/extraction	Schema builder, template library
Matrix	/matrix	Papers × fields table with confidence bars
Analytics	/analytics	Bar charts for methods/year distributions
Clusters	/clusters	Semantic clustering with colored cards
Conflicts	/conflicts	Contradictory findings detection
Zotero	/zotero	Connect, browse collections, sync

---

Key Architecture Decisions

1. Multi-pass extraction — 4-stage pipeline (retrieve → extract → validate → provenance) ensures structured, grounded outputs
2. Hybrid section classifier — Regex heading detection first, LLM fallback for ambiguous chunks
3. Confidence scoring — Multi-factor (source quality, corroboration count, type match, ambiguity detection)
4. FAISS + sklearn fallback — Clustering works with or without FAISS installed
5. Audit logging — Every LLM call is logged with prompt, model, and response

---

How to Run

1. Prerequisites

• Docker and Docker Compose (for PostgreSQL)
• Python 3.11+
• Node.js 20+

2. Start the Database

# From the project root — starts PostgreSQL 15 on port 5433
docker compose up -d

# Verify it's running
docker ps | grep slr_postgres

3. Backend Setup

cd backend

# Create and activate virtual environment
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Initialize Database (create tables & seed default user)
python3 init_db.py

# Run the server
PYTHONPATH=. python3 -m uvicorn app.main:app --reload --host 0.0.0.0 --port 8000
# API docs → http://localhost:8000/docs

4. Frontend Setup

cd frontend
npm install
npm run dev
# Dashboard → http://localhost:3000

---

API Endpoints Summary

Prefix	Endpoints
/api/papers	Upload, list, get, update, delete, process, chunks
/api/zotero	Connect, collections, sync
/api/extraction	Schemas CRUD, templates, run, results, corrections
/api/matrix	Build, get, export (CSV/JSON)
/api/analytics	Methods frequency, year trends, distributions
/api/clusters	Generate, get cached
/api/conflicts	Detect, get cached
/api/health	Health check
/api/overview	Dashboard statistics

Please fill in the env accordingly or add support for a different API for the agents required.