PeerNet++ V3.0

AI-Powered Academic Peer Review System - 2025 SOTA Multi-Provider Stack

Revolutionizes academic paper evaluation through intelligent automation, custom reviewer personalities, plagiarism-first architecture, and blockchain-style audit trails.

🆕 What's New in V3.0

• 🔬 Gemini 2.0 Flash for PDF Vision extraction
• 🤖 Groq Llama 3.1 8B for lightning-fast reviews (560 tok/sec)
• 🧠 Gemini 2.5 Flash Thinking for consensus with reasoning
• 📊 text-embedding-004 for plagiarism detection
• 🔄 Gemma 3 27B Fallback via OpenRouter (FREE!)
• ⚡ Plagiarism-First Architecture - Rejects before wasting API calls
• 📡 Real-time WebSocket Progress with step-by-step live logs

🛠️ Tech Stack

Component	Technology	Purpose
PDF Extraction	Gemini 2.0 Flash Lite	Vision-based text + figure extraction
Embeddings	text-embedding-004	768-dim vectors for similarity
Reviews	Groq Llama 3.1 8B	Fast inference (560 tok/sec)
Consensus	Gemini 2.5 Flash	Built-in reasoning capabilities
Fallback	Gemma 3 27B (OpenRouter)	FREE fallback when rate limited
Database	MongoDB Atlas	Cloud-hosted document store
Real-time	Flask-SocketIO	Live progress updates

🚀 Features

📄 Multi-Modal Paper Upload

• PDF Upload: AI-powered vision extraction with figure descriptions
• JSON Upload: Direct structured metadata input
• API Integration: arXiv, PubMed, Semantic Scholar, OpenAlex

🔍 Plagiarism-First Architecture

• Papers checked BEFORE reviews (saves API costs!)
• 85% similarity threshold auto-rejects duplicates
• Cosine similarity on 768-dim embeddings

🤖 Multi-Provider AI Reviews

• Primary: Groq Llama 3.1 8B (DSPy optimized)
• Fallback 1: Groq direct API
• Fallback 2: Gemma 3 27B via OpenRouter (FREE)
• 5 reviewer personalities: Methodology, Innovation, Communication, Theory, Application

👤 Custom Reviewer Builder

Create personalized AI reviewers with 6 adjustable traits:

• Strictness (Lenient ↔ Harsh)
• Detail Focus (Big Picture ↔ Nitpicky)
• Innovation Bias (Conservative ↔ Novelty-seeking)
• Writing Standards (Relaxed ↔ Perfectionist)
• Methodology Rigor (Flexible ↔ Statistical Purist)
• Optimism (Critical ↔ Encouraging)

📊 Advanced Features

• Real-time Dashboard: Live progress with emojis and step numbers
• Bias Detection: Automated identification and flagging
• Blockchain Ledger: Immutable SHA-256 audit trail
• PDF Report Export: Download complete review reports

🛠️ Quick Start

1. Clone & Setup

   git clone https://github.com/anVSS1/PeerNet.git
   cd PeerNet
   python -m venv venv
   venv\Scripts\activate  # Windows
   # source venv/bin/activate  # Linux/Mac
   pip install -r requirements.txt

2. Configure Environment

   cp .env.example .env
   # Edit .env with your API keys

3. Required API Keys (all have free tiers!)

   • Google AI Studio - Gemini API
   • Groq Console - Groq API
   • OpenRouter - Gemma fallback
   • MongoDB Atlas - Database

4. Run

   python app.py
   # Visit: http://127.0.0.1:5000

⚙️ Configuration

Edit .env file:

# MongoDB Atlas
MONGODB_URI=mongodb+srv://user:pass@cluster.mongodb.net/peernet_plus

# Google Gemini (Vision + Embeddings + Consensus)
GEMINI_API_KEY=your_gemini_key
GEMINI_VISION_MODEL=gemini-2.0-flash-lite
GEMINI_EMBEDDING_MODEL=text-embedding-004
GEMINI_THINKING_MODEL=gemini-2.5-flash

# Groq (Reviews - 560 tokens/sec!)
GROQ_API_KEY=your_groq_key
GROQ_MODEL=llama-3.1-8b-instant

# OpenRouter (Gemma 3 Fallback - FREE!)
OPENROUTER_API_KEY=your_openrouter_key
OPENROUTER_MODEL=google/gemma-3-27b-it:free

# Plagiarism Threshold
PLAGIARISM_SIMILARITY_THRESHOLD=0.85

# Review Settings
MIN_REVIEWERS=3
MAX_REVIEWERS=5

📁 Project Structure

PeerNet++/
├── agents/              # AI Review Agents
│   ├── reviewer_agent.py   # DSPy + Groq + Gemma fallback
│   ├── consensus_agent.py  # Gemini 2.5 Flash Thinking
│   ├── bias_detection_agent.py
│   └── plagiarism_agent.py
├── api/                 # REST API Endpoints
├── dashboard/           # Web Interface & Templates
├── data_collection/     # Paper Intake System
│   ├── pdf_parser.py       # Gemini Vision extraction
│   ├── paper_intake.py     # Plagiarism-first pipeline
│   ├── arxiv_fetcher.py
│   ├── pubmed_fetcher.py
│   └── semantic_fetcher.py
├── models/              # MongoDB Models
├── simulation/          # Review Orchestration
├── utils/               # Utilities & Security
├── app.py               # Main Flask Application
└── requirements.txt

🎯 How It Works

┌─────────────┐    ┌──────────────┐    ┌─────────────┐
│  1. Upload  │───▶│ 2. Extract   │───▶│ 3. Embed    │
│  PDF/JSON   │    │ Gemini Vision│    │ text-emb-004│
└─────────────┘    └──────────────┘    └─────────────┘
                                              │
                                              ▼
┌─────────────┐    ┌──────────────┐    ┌─────────────┐
│ 6. Consensus│◀───│ 5. Reviews   │◀───│4. Plagiarism│
│Gemini 2.5   │    │ Groq/Gemma   │    │ Check FIRST │
└─────────────┘    └──────────────┘    └─────────────┘
       │                                     │
       ▼                                     ▼
┌─────────────┐                       ┌─────────────┐
│ 7. Bias     │                       │  REJECT if  │
│ Detection   │                       │  >85% match │
└─────────────┘                       └─────────────┘

📡 Live Progress Logs

Real-time WebSocket updates show:

🚀 Starting AI review for: Neural Networks for Image...
📋 Step 1/5: Assembling reviewer panel...
👥 3 reviewers assigned: Methodology Expert, Novelty Expert, Clarity Expert
🤖 Step 2/5: Methodology Expert analyzing... (1/3) [Groq Llama 3.1]
🎯 Step 3/5: Building consensus from 3 reviews... [Groq Llama 3.3 70B]
✅ Preliminary decision: Accept (confidence: 85%)
🔍 Step 4/5: Running originality check...
⚖️ Step 5/5: Analyzing for reviewer bias patterns...
🎉 Review complete! ✅ Decision: Accept | 3 reviewers

🤝 Contributing

1. Fork the repository
2. Create feature branch: git checkout -b feature-name
3. Commit changes: git commit -m 'Add feature'
4. Push branch: git push origin feature-name
5. Open Pull Request

📄 License

MIT License - see LICENSE file for details.

👥 Team

• Anass Ouzaouit - GitHub | LinkedIn
• Abdelaziz Afquir - GitHub | LinkedIn
• Mohamed Kannoun - GitHub | LinkedIn

🙏 Acknowledgments

Built with:

• Google Gemini - Vision, Embeddings, Consensus
• Groq - Ultra-fast LLM inference
• OpenRouter - Gemma 3 fallback
• MongoDB Atlas - Cloud database
• Flask - Web framework
• DSPy - LLM optimization

---

Made with ❤️ for the academic community

Version: 3.0.0 | Last Updated: December 2025