NLP: Resume Classification System

Overview

Resume-Analysis-NLP is a full-stack application that uses Natural Language Processing (NLP) and Deep Learning to classify resumes into job categories. It provides a modern web interface for users to upload resumes (PDF, DOCX, DOC) or paste text, and instantly receive a predicted job category and confidence score.

• Backend: FastAPI, TensorFlow/Keras, NLTK, joblib
• Frontend: HTML, CSS, JavaScript (vanilla)
• Deployment: Docker-ready

---

Features

• Classifies resumes into 25+ job categories using a trained deep learning model.
• Accepts input as either:
  • Uploaded file (.pdf, .docx, .doc)
  • Pasted text
• Displays results: Category and confidence score.
• Preview: Shows extracted text from uploaded files.
• Retraining endpoint (for advanced users).
• Modern, responsive frontend.

---

Directory Structure

Resume-Analysis-NLP/
│
├── main.py                  # FastAPI backend (API, model loading, endpoints)
├── requirements.txt         # Python dependencies
├── Dockerfile               # For containerized deployment
│
├── src/                     # Core backend modules
│   ├── model/               # Model architecture, saved model, tokenizer, encoder
│   ├── utils/               # Utilities (logging, helpers, file extraction)
│   ├── training/            # Training pipeline and scripts
│   ├── preprocessing/       # Data preprocessing logic
│   └── inference/           # Inference and prediction logic
│
├── dataset/                 # Resume datasets (CSV, sample PDFs)
│
├── frontend/                # Frontend (static files)
│   ├── index.html           # Main UI
│   ├── styles.css           # Styling
│   └── script.js            # Interactivity/API calls
│
├── logs/                    # Log files
├── tests/                   # Test scripts

---

Setup & Installation

1. Clone the repository

git clone <repo-url>
cd Resume-Analysis-NLP

2. Install dependencies

Recommended: Use a virtual environment.

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install --upgrade pip
pip install -r requirements.txt

3. Download/prepare model files

Ensure the following files exist in src/model/:

• best_model.keras
• trained_tokenizer.json
• OHEncoder.joblib

If not, retrain the model using the provided notebook or scripts.

4. Run the app

uvicorn main:app --reload

Visit http://localhost:8000 in your browser.

---

Docker Deployment

Build and run the app in a container:

docker build -t resume-analyser .
docker run -p 8000:8000 resume-analyser

---

API Endpoints

POST /classify_resume/text/

• Input: JSON: { "resume_text": "..." }
• Output: Category, confidence, extracted text

POST /classify_resume/file/

• Input: Form-data: file (PDF, DOCX, DOC)
• Output: Category, confidence, extracted text

POST /classify_resume/train/

• Input: (Advanced) Triggers retraining (see code for details)

GET /

• Frontend UI (index.html)

GET /logs/status

• Returns logging status

---

Frontend Usage

• Paste resume text or upload a file in the left column.
• Click Classify Resume.
• See the predicted category and confidence below.
• The right column shows the uploaded file name and extracted text.

---

Dataset

• Place your resume datasets (CSV, PDF) in the dataset/ directory.
• Example files: resume_new.csv, resume_dataset.csv, DummyResume.pdf

---

Model & Training

• Model is a custom Keras text classifier.
• Preprocessing uses NLTK, custom tokenization, and one-hot encoding.
• Training scripts and logic are in src/training/ and src/preprocessing/.
• See NLP_Resume_Classification.ipynb for EDA and prototyping.

---

Customization

• Add new categories: Update your dataset and retrain the model.
• Change model architecture: Edit src/model/model.py.
• Logging: Configured in src/utils/logger.py.

---

Contributing

1. Fork the repo
2. Create a feature branch
3. Commit your changes
4. Open a pull request

---

License

MIT License.

---

Acknowledgements

• Inspired by open datasets and NLP research.
• Built with FastAPI, TensorFlow, and NLTK.

---

For more details, see the code and comments in each module.