iWork Job Portal

iWork is a job portal built using HTML, CSS, and Django, featuring a cutting-edge AI-powered job recommender system. The recommender, HGCNL-JRec, leverages both Natural Language Processing (NLP) and Graph Neural Networks (GNN) to provide highly relevant job recommendations to users based on their resumes and job interactions.

---

🚀 Key Features

• User authentication and profile management (Employer/Employee roles)
• Job postings, search, and application workflows
• AI-powered job recommendation system (HGCNL-JRec)
• Resume parsing and skill extraction using NLP
• Admin dashboard (Django admin)
• Responsive web UI with custom templates

---

🧠 AI Recommender System: HGCNL-JRec

Overview

The heart of this portal is the HGCNL-JRec recommender, a hybrid model that combines:

• LDA (Latent Dirichlet Allocation) for topic modeling (via gensim)
• Graph Neural Networks (GNN) for learning from user-job-application relationships (via PyTorch)
• spaCy with custom entity rulers for extracting skills, degrees, and majors from resumes and job descriptions

How It Works

1. Resume & Job Parsing:

   • When an employee uploads a PDF resume, it is parsed using pdfminer and cleaned with NLP techniques.
   • Skills, degrees, and majors are extracted using spaCy and custom entity rulers (website/recommend/recFiles/).
   • Job descriptions are similarly processed.

2. Feature Embedding:

   • Extracted features are embedded into a unified representation.
   • LDA assigns a topic to each resume/job for semantic grouping.

3. Graph Construction:

   • Users, jobs, and their interactions (applications) are represented as nodes and edges in a heterogeneous graph.
   • The graph is stored and updated in website/recommend/recFiles/web_graph.pt.

4. Recommendation:

   • The GNN model (HGCNL-JRec) learns from the graph structure and content features.
   • For a given user, the model recommends jobs they are most likely to be interested in, excluding those already applied to.
   • The top-K recommendations are returned and displayed in the UI.

Code & Artifacts

• Model code: website/recommend/recommeder.py, graph.py, preprocess.py
• Pretrained models and rulers: website/recommend/recFiles/
• Global variables: website/recommend/global_vars.py (paths to models, rulers, and graph)

Dependencies

• torch, gensim, spacy, pdfminer, nltk, and more (see requirements.txt)
• Pretrained spaCy entity rulers and LDA models are included in the repo for immediate use

Customization & Extension

• You can retrain the LDA or GNN models with your own data
• Entity rulers for skills, degrees, and majors can be extended for new domains
• The recommender logic is modular and can be adapted for other recommendation tasks

---

🗂️ Project Structure

• website/ - Main Django app (views, models, forms, templates, static files)
• website/recommend/ - Recommender system code and models
• employee_resumes/ - Uploaded resumes (PDF)
• Notebooks/ - Jupyter notebooks for data processing and experiments
• job_portal/ - Django project settings and configuration
• requirements.txt - All dependencies (Django, PyTorch, gensim, spaCy, etc.)

---

🛠️ Installation & Setup

1. Clone the repository:

   git clone https://github.com/benbarekfatima/JobPortal.git
   cd JobPortal

2. Create and activate a virtual environment:

   python -m venv job_env
   # On Windows:
   job_env\Scripts\activate
   # On Unix/Mac:
   source job_env/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. Download NLTK stopwords (if not auto-downloaded):

   python -c "import nltk; nltk.download('stopwords')"

5. Apply migrations and create a superuser:

   python manage.py migrate
   python manage.py createsuperuser

6. Run the development server:

   python manage.py runserver

7. Access the app:
   Open http://127.0.0.1:8000/ in your browser.

---

🌐 Deployment

• Vercel: Configured via vercel.json to deploy using Python 3.9 and the Django WSGI entry point.

---

📄 Notes

• Resume Uploads: Only PDF files are accepted.
• Skills/majors/degrees extraction: Uses custom spaCy entity rulers (see website/recommend/recFiles/).
• Admin: Access Django admin at /admin/ after creating a superuser.
• Customization: You can extend the recommender, add new features, or modify the UI via the templates.

---

🤝 Contributing

Pull requests are welcome! For major changes, please open an issue first to discuss what you would like to change.