OriginLytics

AI-powered content authenticity and intelligence platform. 10 analysis tools, 17+ metrics, real ML models — not just API wrappers.

🏆 Top Project — ELVTR AI Architecture Program (2025)

Features

Analysis Tools (10)

• AI Detector — RoBERTa classifier + DistilGPT-2 perplexity with ensemble scoring
• Paraphraser — Content rewriting with AI detection awareness
• Summarizer — GPT-4 powered content summarization
• Grammar Checker — Writing quality analysis
• Citation Generator — Automated citation formatting
• Readability Checker — 9 readability algorithms via textstat
• Essay Checker — Academic writing evaluation
• Text Comparison — Side-by-side content diff
• Plagiarism Checker — Content originality verification
• Word Counter — Text statistics and metrics

Analysis Metrics (17+)

• AI Detection — RoBERTa-based classification + perplexity scoring via DistilGPT-2
• GEO (Generative Engine Optimization) — 8 weighted metrics for AI search visibility (citation structure, source credibility, structured data, content freshness, author attribution, factual clarity, data presence, content depth)
• SEO Analysis — Technical SEO audit with actionable recommendations
• Toxicity Detection — Content safety via Detoxify (Unitary) models
• Emotional Analysis — NRC Emotion Lexicon sentiment scoring
• Accessibility — WCAG compliance checking
• Premium Metrics — Readability (9 algorithms), linguistic complexity (POS tagging, NER via NLTK), statistical writing fingerprint
• Ensemble Scoring — Weighted normalization across all signals

Architecture

originlytics/
├── src/                          # React/TypeScript frontend (Vite)
│   ├── pages/                    # 10 tool pages + Dashboard
│   ├── components/               # GEO, SEO, Accessibility, Emotional analysis panels
│   ├── scoring/                  # Ensemble scoring + normalization
│   ├── analysis.js               # Core backend — Python subprocess orchestration
│   ├── server.js                 # Express API server
│   └── ...
├── python/                       # ML analysis pipeline
│   ├── ai_detector.py            # RoBERTa classifier + DistilGPT-2 perplexity
│   ├── ai_detector_quick.py      # Fast detection mode
│   ├── geo_analyzer.py           # GEO scoring (8 metrics, weighted)
│   ├── seo_analyzer.py           # SEO evaluation
│   ├── premium_metrics.py        # Readability, linguistics, statistical fingerprint
│   ├── emotion_analyzer.py       # NRC Emotion Lexicon analysis
│   ├── metrics.py                # Base metrics utilities
│   └── requirements.txt          # Python dependencies
├── server.js                     # Express entry point
├── website/                      # Marketing site (originlytics.com)
└── database/                     # Migration scripts

Models

Model	Purpose	Source
Hello-SimpleAI/chatgpt-detector-roberta	AI content classification	HuggingFace
distilgpt2	Perplexity-based AI detection	HuggingFace
detoxify (Unitary)	Toxicity scoring	PyPI
gpt-4.1	Summarization, key points, translation	OpenAI API
NRC Emotion Lexicon	Emotional tone analysis	NRC
NLTK (punkt, POS tagger, NER)	Linguistic complexity	NLTK
textstat	Readability (9 algorithms)	PyPI

Quick Start

Prerequisites

• Node.js 18+
• Python 3.10+
• An OpenAI API key

Setup

git clone https://github.com/GixGosu/Originlytics.git
cd Originlytics

# Node dependencies
npm install

# Python dependencies
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

# NLTK data
python -c "import nltk; nltk.download('punkt'); nltk.download('averaged_perceptron_tagger'); nltk.download('maxent_ne_chunker'); nltk.download('words')"

# Environment
cp .env.example .env
# Add your OPENAI_API_KEY to .env

# Run
node server.js          # API on :8080
npm run dev             # Frontend on :5173

Environment Variables

Variable	Required	Description
OPENAI_API_KEY	Yes	OpenAI API key for GPT-4 and moderation
SUPABASE_URL	No	Supabase project URL (auth/tokens)
SUPABASE_ANON_KEY	No	Supabase anon key
STRIPE_SECRET_KEY	No	Stripe key (premium tier billing)

How It Works

1. Input → URL or raw text submitted via React frontend
2. Extraction → Playwright renders page, extracts clean text
3. Parallel Analysis Pipeline:
   • Python subprocess: RoBERTa AI detection + perplexity scoring
   • Python subprocess: GEO analysis (8 weighted metrics)
   • Python subprocess: SEO evaluation
   • Python subprocess: Toxicity via Detoxify
   • Python subprocess: Emotional analysis via NRC Lexicon
   • Python subprocess: Premium metrics (readability, linguistics, fingerprint)
   • Node.js: Accessibility analysis
   • OpenAI API: Summary, key points, translation
4. Ensemble Scoring → Weighted normalization across all signals
5. Dashboard → Results with gauge rings, metric breakdowns, filtering, CSV export, PDF export

Tech Stack

• Frontend: React 19, TypeScript, Vite, CSS design system
• Backend: Express, Node.js, Python subprocess orchestration
• ML: PyTorch, HuggingFace Transformers, NLTK, textstat, scipy, Detoxify
• APIs: OpenAI GPT-4.1, OpenAI Moderation
• Auth: Supabase (optional)
• Payments: Stripe (optional, for premium tier)
• Scraping: Playwright

Contributing

Pull requests welcome. For major changes, open an issue first.

License

MIT

---

Built by BrineShrimp Games