🧠 Internship Detection WebApp

Detecting Genuine vs. Fake Internships Using Machine Learning

📋 Project Overview

The Internship Detection WebApp is an AI-powered solution that helps students verify whether internship postings are genuine or fake.
It leverages Complemented Naive Bayes (CNB) for text classification and provides a Streamlit-based interface for instant prediction.

This project aims to empower students by reducing fraud and misinformation during internship searches.

🚨 Problem Statement

The rise of fake internship offers has become a major concern in the digital era. Many fraudulent listings exploit students’ ambitions, leading to:

• Wasted time and effort
• Loss of trust in legitimate platforms
• Potential personal and financial risks

Thus, there is an urgent need for an automated verification system to ensure the authenticity of internship postings.

🎯 Objective

To develop a machine learning model that accurately classifies internship postings as Genuine or Fake, providing:

• Instant predictions
• High precision and recall
• User-friendly web interface for real-time verification

🧩 Features

• 🧾 Paste internship text: Users input full internship details.
• ⚙️ Instant prediction: Model outputs “Genuine” or “Fake” with confidence.
• 🎨 Simple Streamlit UI: Interactive and easy-to-use design.
• 📊 Model Accuracy: Achieved 97% accuracy with robust validation.

🧠 Machine Learning Model

Algorithm Used:

Complemented Naive Bayes (CNB)
Chosen for its strong performance on text classification tasks and ability to handle imbalanced data.

Key Advantages:

• Reduces bias towards frequent classes.
• Performs better than Multinomial Naive Bayes on sparse data.
• Offers high precision and recall, minimizing false positives/negatives.

Data Processing:

• Dataset:
  • Real internship data scraped from Internshala using BeautifulSoup.
  • Fake internship data sourced from Kaggle.
• Text Preprocessing:
  • Tokenization, stopword removal, and TF-IDF vectorization.
• Feature Engineering:
  • Extracted key fields (company name, description, etc.).
• Data Balancing:
  • Ensured equal representation of fake and genuine classes.

🛠️ Tech Stack

Component	Technology
Programming Language	Python
Framework	Streamlit
ML Library	Scikit-learn
Model Serialization	Joblib
Data Collection	BeautifulSoup, Kaggle datasets
Version Control	GitHub

💻 Installation & Setup

1. Clone this repository

   git clone https://github.com/Fatimibee/internship_detection
   cd internship-detection

2. Run the Streamlit App

   streamlit run page.py

3. Paste internship text into the input box and view prediction instantly!

📈 Model Performance

Metric	Score
Accuracy	97%
Fake Precision	96%
Fake Recall	94%

The model generalizes well and was validated on unseen data to ensure real-world applicability.

🚀 Future Enhancements

• 🔍 Advanced NLP models – Integrate transformers (e.g., BERT) for deeper semantic understanding.
• ☁️ Cloud Deployment – Deploy for public access (e.g., AWS, Streamlit Cloud).
• 📱 Mobile App – Allow internship verification on the go.
• 🗂️ User Feedback Loop – Improve accuracy based on real user responses.

🧾 Project Contributors

Developed by:
👩‍💻 Fatimi Bee
Roll No.: 23ESKCS077
B.Tech CSE, 5th Semester
Swami Keshvanand Institute of Technology, Management & Gramothan, Jaipur

📚 References

• Internshala dataset (scraped data)
• Kaggle: Fake Internship Listings Dataset
• Scikit-learn Documentation
• Streamlit Official Docs