Fake News Detector in Python - Build Your Own Truth Checker Tool

Fake News Detector in Python – Build Your Own Truth Checker Tool

Today, misinformation doesn’t stay limited to social platforms; it spreads across blogs, websites, and messaging apps.

It affects elections, markets, public opinion, and even personal relationships. As developers, we don’t just consume technology — we create solutions.

In this article, I’ll show you how to build a Fake News Detector using Python.

No heavy math, no boring theory — just practical learning.

Why Fake News Detection Matters Today

Every day:

• WhatsApp forwards spread misinformation
• Fake headlines manipulate clicks
• AI-generated content increases confusion

Manually verifying news is slow and unreliable.

This is where Machine Learning helps.

A Fake News Detector:

• Reads news text
• Learns language patterns
• Predicts whether news is Real or Fake

How a Fake News Detector Works (Simple Explanation)

Think of it like spam detection for news.

Step-by-step logic:

• Input → News article text
• Processing → Clean & understand text
• Learning → Compare with past examples
• Prediction → Fake or Real

The computer doesn’t “understand” truth —

it recognizes patterns.

Tools & Libraries Used

We’ll use industry-standard Python libraries:

• pandas – data handling
• numpy – numerical operations
• sklearn – machine learning
• TfidfVectorizer – text understanding
• PassiveAggressiveClassifier – fast & effective model

These are safe, trusted, and AdSense-friendly topics.

Dataset for Fake News Detection

A dataset usually contains:

• text → news content
• label → REAL or FAKE

Example:

Text	Label
Government announces new policy	REAL
Celebrity cured cancer with lemon	FAKE

(You can use Kaggle Fake News datasets or your own collected data.)

Step 1: Import Required Libraries

Install Required Libraries

python3 -m pip install scikit-learn pandas

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import PassiveAggressiveClassifier
from sklearn.metrics import accuracy_score

Step 2: Load the Dataset

data = pd.read_csv("news.csv")
print(data.head())

Our dataset should contain:

text

label

Step 3: Split Data into Training & Testing

X = data['text']
y = data['label']

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

Why split?

• Train → learning
• Test → real-world checking

Step 4: Convert Text into Numbers (TF-IDF)

Computers don’t understand words — they understand numbers.

vectorizer = TfidfVectorizer(stop_words='english', max_df=0.7)
X_train_tfidf = vectorizer.fit_transform(X_train)
X_test_tfidf = vectorizer.transform(X_test)

TF-IDF:

• Ignores common words (is, the, and)
• Focuses on meaningful terms

Step 5: Train the Fake News Model

model = PassiveAggressiveClassifier(max_iter=50)
model.fit(X_train_tfidf, y_train)

Why PassiveAggressive?

• Fast
• Good for text data
• Used in real NLP systems

Step 6: Test the Model Accuracy

y_pred = model.predict(X_test_tfidf)
accuracy = accuracy_score(y_test, y_pred)
print("Accuracy:", accuracy)

Typical accuracy:

• 90%+ with a clean dataset

Step 7: Detect Fake News from User Input

def detect_news(news_text):
    transformed = vectorizer.transform([news_text])
    prediction = model.predict(transformed)
    return prediction[0]
news = "Breaking: Scientists confirm water cures all diseases"
print(detect_news(news))

Output:

FAKE

Real-World Use Cases

Fake News Detector can be used in:

• News websites
• Browser extensions
• Social media filters
• WhatsApp message verification
• Journalism tools

This is not just a demo project, but a portfolio-ready idea.

Source Code:

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import SGDClassifier
from sklearn.metrics import accuracy_score

# 1. Load dataset
data = pd.read_csv("news.csv")

# Check required columns
if 'text' not in data.columns or 'label' not in data.columns:
    raise ValueError("Dataset must contain 'text' and 'label' columns")

# 2. Split data
X = data['text']
y = data['label']

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

# 3. Text vectorization (TF-IDF)
vectorizer = TfidfVectorizer(
    stop_words='english',
    max_df=0.7
)

X_train_tfidf = vectorizer.fit_transform(X_train)
X_test_tfidf = vectorizer.transform(X_test)

# 4. Train model (Passive-Aggressive via SGDClassifier)
model = SGDClassifier(
    loss='hinge',              # same as Passive-Aggressive
    penalty=None,
    learning_rate='pa1',
    eta0=1.0,
    max_iter=1000,
    random_state=42
)

model.fit(X_train_tfidf, y_train)

# 5. Evaluate model
y_pred = model.predict(X_test_tfidf)
accuracy = accuracy_score(y_test, y_pred)

print("✅ Model trained successfully")
print(f"🎯 Accuracy: {accuracy * 100:.2f}%")

# 6. Prediction function
def detect_fake_news(news_text):
    text_vector = vectorizer.transform([news_text])
    prediction = model.predict(text_vector)
    return prediction[0]

# 7. User input loop
print("\n📰 Fake News Detector (type 'exit' to stop)\n")

while True:
    user_input = input("Enter news text: ")

    if user_input.lower() == "exit":
        print("👋 Exiting Fake News Detector")
        break

    result = detect_fake_news(user_input)
    print("🔍 Prediction:", result)
    print("-" * 50)

Result:

Frequently Asked Questions (FAQ)

Is Fake News Detection legal?

Yes, if used for educational and research purposes.

Can beginners build this?

Absolutely. Basic Python knowledge is enough.

Is this project good for resumes?

Yes, especially for:

• Data Science
• Python Developer
• Machine Learning roles

Final Thoughts

Fake news is a serious digital problem, but technology can fight it.

By building a Fake News Detector:

• You learn real Machine Learning
• You create meaningful software

If you are serious about Python and AI —

this project deserves a place on your skill and resume.