⚠️ Hiyari-Hatto (Near-Miss) AI Analyzer

An NLP-driven framework for classifying unstructured industrial safety reports.

This project automates the risk assessment of "Hiyari-Hatto" (Near-Miss) incidents in construction environments. By using Natural Language Processing (NLP) and rule-based logic, it converts unstructured worker reports into structured safety data, enabling site managers to predict and prevent accidents before they occur.

---

📸 Dashboard Preview

(Figure 1: Real-time analysis interface showing risk classification and keyword extraction)

---

🚀 Key Features

• Natural Language Processing (NLP):
  • Analyzes raw text using TextBlob and NLTK.
  • Extracts safety-critical keywords (e.g., "leak," "spark," "voltage").
• Automated Risk Classification:
  • Categorizes incidents into High (Critical), Medium (Caution), and Low (Monitor) based on industry safety standards.
• Real-Time Visualization:
  • Interactive dashboard built with Flask and Bootstrap 5.
  • Visualizes risk trends using Chart.js.
• Industrial Applicability:
  • Designed to handle noisy, non-technical language often found in on-site worker logs.

---

🛠️ Tech Stack

Component	Technology
Backend	Python, Flask (Web Server)
NLP Engine	NLTK, TextBlob, Pandas
Frontend	HTML5, CSS3, Bootstrap 5
Data Analysis	Scikit-learn, NumPy

---

📂 Project Structure

AI_Hazard_Risk_Analyzer/
├── data/                  # Sample Hiyari-Hatto datasets (CSV)
├── src/
│   ├── app.py             # Main Flask application logic
│   └── templates/         # UI Dashboard (HTML/Jinja2)
├── requirements.txt       # Research dependencies
└── README.md              # Project documentation

---

⚡ How to Run

1. Setup Environment

# Create Virtual Environment (Sandbox)
python -m venv venv

# Activate (Windows)
.\venv\Scripts\activate

# Install Dependencies
pip install -r requirements.txt

2. Initialize NLP Resources

# Download required NLTK data (run once)
python -m textblob.download_corpora

3. Launch Application

# Navigate to the Source Folder
cd src

# Start the Analysis Engine
python app.py

Access the dashboard at: http://127.0.0.1:5001

---

🔧 Troubleshooting (Windows Users)

If you see a "running scripts is disabled" error when trying to activate the environment, run this command in PowerShell to allow script execution:

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope Process

Then try running .\venv\Scripts\activate again.

---

🔬 Research Context

This tool was developed to address the "Data Silo" problem in industrial safety. While large construction firms collect thousands of Near-Miss reports, they are often stored as text logs that are difficult to analyze at scale.

Future Scope:

1. Integration with Formal Verification methods to model safety state transitions.
2. Expansion of the dataset to include multi-lingual support (Hindi/Japanese).

---

👤 Author

Gaurav Dev

• Research Interests: Industrial Safety Systems, Formal Methods, NLP.