CredCheck AI: Fake-news classifier built with TF-IDF & logistic regression

ROLEML Engineer
DURATIONJul 2025 - Sept 2025
TEAM1 ML Engineer
TECH STACKPython, NLP, scikit-learn, TF-IDF, Flask

Overview

CredCheck AI is an AI-powered online news credibility detector designed to help users identify potentially fake or misleading news. It provides a simple way for students, employees, researchers, and everyday internet users to assess news content before sharing or relying on it. By analyzing news headlines and article text, the system classifies content as potentially credible or fake, helping users become more aware of misinformation online.

Problem

The widespread sharing of misinformation online makes it difficult for people to determine whether news articles are trustworthy. Manually checking the credibility of news can be time-consuming, especially for users who are unfamiliar with fact-checking tools and techniques.

  • Difficulty Identifying Fake News: Users may struggle to distinguish reliable news articles from fabricated or misleading information.
  • Time-Consuming Fact-Checking: Verifying news often requires searching multiple sources and comparing information before deciding whether an article is trustworthy.
  • Limited Technical Knowledge: Some users may not be familiar with available fact-checking methods, making it harder to evaluate suspicious news content.
  • Rapid Spread of Misinformation: Misleading articles can spread quickly through online platforms, influencing readers before the information is verified.

Solution

CredCheck AI is a web-based news classification system that uses machine learning and text processing to assess the credibility of news content. It applies a Logistic Regression model within a machine learning pipeline that uses text vectorization to convert written content into numerical data that the model can analyze.

CredCheck AI - News Credibility Checker Interface and Prediction ResultClick to zoom

The model was trained on a cleaned dataset containing approximately 40,000 records and achieved a reported accuracy of 98% during evaluation. Users can check news by copying the headline and article body into the system's input fields and clicking the check button to receive a classification result.

CredCheck AI - Confusion Matrix (Real vs Fake News Classification)Click to zoom

With its simple interface and automated text analysis, CredCheck AI makes machine learning-based news classification more accessible to users without requiring specialized technical knowledge. The tool is intended to support users in evaluating online information, rather than replace independent fact-checking or verification through reliable sources.

CredCheck AI - Evaluation Metrics by Class (Precision, Recall, F1-Score)Click to zoom