Fake News Detection App
This project delivers an end-to-end NLP solution for identifying misinformation by classifying news articles as real or fake. Using a dataset of labeled news stories from Kaggle, the system preprocesses text, extracts features with TF-IDF vectorization, and trains a Decision Tree Classifier for binary classification. The final model is deployed as an interactive Streamlit application, enabling real-time predictions through a simple web interface.
Year
2025
Service
ML Model
Category
NLP
Tool
Streamlit
Description:
This project focuses on detecting misinformation by classifying news articles as either real or fake using Natural Language Processing (NLP) and machine learning. The workflow begins with combining article titles and content into a single text feature, followed by extensive preprocessing steps such as lowercasing, punctuation removal, stopword removal, and lemmatization using NLTK.
The preprocessed text is transformed into numerical features using TF-IDF vectorization (limited to 5,000 features for efficiency). A Decision Tree Classifier is trained to perform binary classification, and both the model and vectorizer are serialized using Pickle for deployment.
The solution is implemented as an interactive Streamlit web application, where users can input a news article and instantly receive a prediction along with an intuitive interface. The project demonstrates the complete lifecycle of an NLP application: from dataset preparation and text preprocessing to model training, evaluation, and deployment for real-time use.
Key Highlights:
Problem: Combat misinformation by automatically flagging fake news.
Approach: TF-IDF + Decision Tree Classifier.
Dataset: Fake and Real News Dataset from Kaggle.
Deployment: Streamlit app for instant prediction.
Tools:
Python
Pandas
scikit-learn
NLTK
TfidfVectorizer
Streamlit
Pickle
Streamlit
