🧠 AI Intelligence Platform — Java ML System

An end-to-end Machine Learning platform combining fraud detection, regression, NLP, recommendation systems, and deep learning into a unified Java system with production-ready REST APIs.

---

📌 About

This project is a full-stack Java ML platform that brings together five distinct machine learning domains into a single, cohesive system. Each module is independently trained, served through its own REST endpoint, and designed with graceful fallbacks — making the platform resilient, extensible, and production-aware.

---

🏗️ Architecture

ai-intelligence-platform-java-ml/
│
├── src/main/java/com/mazen/aiplatform/
│   ├── Application.java                          # Spring Boot entry point
│   │
│   ├── controllers/
│   │   ├── FraudController.java                  # POST /api/fraud/predict
│   │   ├── PriceController.java                  # POST /api/price/predict
│   │   ├── NlpController.java                    # POST /api/nlp/sentiment & /spam
│   │   ├── RecommendController.java              # POST /api/recommend
│   │   ├── ImageController.java                  # POST /api/image/predict
│   │   └── GlobalExceptionHandler.java           # Unified error handling
│   │
│   ├── services/
│   │   ├── FraudService.java                     # Fraud detection orchestration
│   │   ├── PriceService.java                     # Price prediction orchestration
│   │   ├── NlpService.java                       # Sentiment + Spam orchestration
│   │   ├── RecommendService.java                 # Recommendation orchestration
│   │   └── ImageService.java                     # Image classification orchestration
│   │
│   ├── ml/
│   │   ├── fraud/FraudDetector.java              # Weka J48 + rule-based fallback
│   │   ├── regression/PricePredictor.java         # Gradient descent regression
│   │   ├── nlp/SentimentAnalyzer.java            # Lexicon-based sentiment engine
│   │   ├── nlp/SpamDetector.java                 # Keyword + pattern spam scorer
│   │   ├── recommender/RecommenderEngine.java    # Collaborative filtering (cosine)
│   │   └── deep_learning/ImageClassifier.java    # DL4J 4-layer dense network
│   │
│   ├── models/
│   │   ├── FraudRequest.java
│   │   ├── PriceRequest.java
│   │   ├── NlpRequest.java
│   │   ├── RecommendRequest.java
│   │   └── PredictionResponse.java
│   │
│   └── utils/
│       └── CsvUtils.java                         # CSV parsing utility
│
├── datasets/                                      # Auto-loaded on startup
├── saved_models/                                  # PMML / ONNX exports
├── pom.xml
└── README.md

---

🧠 ML Modules Overview

Module	Algorithm	Library	Fallback
Fraud Detection	J48 Decision Tree	Weka	Rule-based risk scoring engine
Price Prediction	Linear Regression (Gradient Descent)	Custom	Weighted estimation formula
Sentiment Analysis	Lexicon-based + Negation Handling	Custom	Seed lexicon (45+ words)
Spam Detection	Keyword + Pattern Scoring	Custom	Keyword matching
Recommendations	Collaborative Filtering (Cosine Similarity)	Custom	Empty response for unknown users
Image Classification	Dense Neural Network (784→256→128→64→10)	DL4J	Architecture-only (requires MNIST training)

---

⚙️ How It Works

 ┌───────────┐     ┌──────────────┐     ┌───────────────┐     ┌──────────────┐
 │  Request   │ ──▶ │  Controller  │ ──▶ │    Service    │ ──▶ │  ML Module   │
 │ (JSON API) │     │ (Validation) │     │ (Orchestrate) │     │ (Predict)    │
 └───────────┘     └──────────────┘     └───────────────┘     └──────────────┘
                                                                      │
                                                                      ▼
                                                              ┌──────────────┐
                                                              │   Response   │
                                                              │  (JSON + %)  │
                                                              └──────────────┘

1. Receive a JSON request through the REST API
2. Validate input using Bean Validation annotations
3. Route to the appropriate service and ML module
4. Predict using the trained model (or graceful fallback)
5. Return a structured response with prediction, confidence, and metadata

---

▶️ Quick Start

Prerequisites: Java 17+, Maven

# Clone
git clone https://github.com/mazen-naji/ai-intelligence-platform-java-ml.git
cd ai-intelligence-platform-java-ml

# Build
mvn clean install

# Run
mvn spring-boot:run

The server starts at http://localhost:8080.

---

📡 API Examples

Fraud Detection

curl -X POST http://localhost:8080/api/fraud/predict \
  -H "Content-Type: application/json" \
  -d '{"amount": 15000, "transactionType": "online"}'

Price Prediction

curl -X POST http://localhost:8080/api/price/predict \
  -H "Content-Type: application/json" \
  -d '{"size": 120, "rooms": 3, "location": "Beirut", "age": 10}'

Sentiment Analysis

curl -X POST http://localhost:8080/api/nlp/sentiment \
  -H "Content-Type: application/json" \
  -d '{"text": "This product is amazing!"}'

Spam Detection

curl -X POST http://localhost:8080/api/nlp/spam \
  -H "Content-Type: application/json" \
  -d '{"text": "WIN FREE MONEY NOW!!!"}'

Recommendations

curl -X POST http://localhost:8080/api/recommend \
  -H "Content-Type: application/json" \
  -d '{"userId": 1, "maxResults": 5}'

---

📊 Datasets

All datasets are inside /datasets and loaded automatically on startup.

File	Records	Purpose
fraud_transactions.csv	100	Transaction classification (fraud / legitimate)
house_prices.csv	100	Real estate price regression
sentiment_data.csv	100	Sentiment classification (positive / negative / neutral)
spam_emails.csv	100	Email classification (spam / ham)
recommendations.csv	100	User-item ratings for collaborative filtering

---

🧪 Testing

mvn test

Unit tests cover all ML modules: fraud detection, sentiment analysis, spam detection, price prediction, and the recommendation engine.

---

🔌 Extensibility

Adding a new ML module requires zero changes to existing code — just follow the established pattern:

// 1. Create the ML module
@Component
public class AnomalyDetector {
    public Map<String, Object> detect(double[] features) {
        // Your anomaly detection logic
    }
}

// 2. Create the service
@Service
public class AnomalyService {
    private final AnomalyDetector detector;
    // Orchestration logic
}

// 3. Create the controller
@RestController
@RequestMapping("/api/anomaly")
public class AnomalyController {
    @PostMapping("/detect")
    public ResponseEntity<PredictionResponse> detect(@RequestBody ...) {
        // Expose via REST
    }
}

---

🗺️ Roadmap

• Docker containerization
• React frontend dashboard
• Real-time streaming with Apache Kafka
• Cloud deployment (AWS / Azure)
• JWT authentication
• Model versioning and A/B testing
• Swagger / OpenAPI documentation

---

🛠️ Tech Stack

Language	Java 17+
Framework	Spring Boot 4.0
ML Libraries	Weka, Tribuo, Deeplearning4j, Apache OpenNLP
Build Tool	Maven
Model Export	PMML (Weka / Tribuo), ONNX (Deep Learning)
Testing	JUnit 5, Spring Boot Test

---

🤝 Contributing

Contributions are welcome! Feel free to fork and submit a pull request.

1. Fork the repository
2. Create your feature branch — git checkout -b feature/new-module
3. Commit your changes — git commit -m "Add anomaly detection module"
4. Push to the branch — git push origin feature/new-module
5. Open a Pull Request

---

Built by Mazen Naji

⭐ If you found this useful, a star goes a long way!