Online Fake News Detection System

Abstract

The rapid proliferation of fake news and misleading information on digital platforms poses a significant threat to public trust, social stability, and informed decision-making. Existing moderation systems relying on manual review or static keyword-based filtering are inadequate in addressing the scale and velocity of online misinformation. Theonline fake news detection system designed to automatically detect abnormal or misleading textual information. The proposed system employs Natural Language Processing (NLP) techniques for text preprocessing and feature extraction, combined with machine learning classifiers—specifically Logistic Regression and Naive Bayes—to categorize content as Normal or Abnormal. Additionally, a risk-level analysis module assigns severity ratings (Low, Medium, High) to flagged content, enabling prioritized moderator intervention. The system is deployed as a web-based application using Flask. Experimental evaluation demonstrates competitive classification accuracy, reduced moderation latency, and practical utility as an early warning tool for content moderators.

Introduction

The text describes an AI-based fake news detection system designed to address the growing problem of misinformation on digital platforms. Traditional methods like manual review and keyword filtering are insufficient for handling large volumes of online content.

The proposed system uses Natural Language Processing (NLP) for text preprocessing and feature extraction, along with machine learning models such as Logistic Regression and Naive Bayes to classify content as normal or abnormal. It also includes a risk-level module that categorizes misinformation into low, medium, and high severity to help prioritize moderation.

The system is implemented as a Flask-based web application and serves as an automated tool to detect fake or misleading news. Experimental results show good accuracy, faster detection, and reduced moderation time, making it an effective early warning system for content moderation.

Conclusion

This paper presented an AI-based early warning and moderation support system for detecting abnormal and misleading textual information. The system integrates NLP-driven preprocessing, TF-IDF feature extraction, and machine learning classification within a Flask-based web application. Logistic Regression achieved a classification accuracy of 92.4% on the test dataset, demonstrating significant improvement over traditional rule-based filtering methods. The risk-level analysis module further enhances the practical utility of the system by enabling moderators to prioritize high-risk content. The proposed system provides a scalable, maintainable, and extensible foundation for automated content moderation in online environments.

References

[1] KUN WANG , YUZHEN YANG ,ANDXIAOYANG WANG\"Spectral Clustering-Guided News Environments Perception for Fake News Detection\" -23 December 2024, DOI-0.1109/ACCESS.2024.3521015 [2] AHMED HASHIM JAWAD ALMARASHY, PEDRAM SALEHPOUR \"Enhancing Fake News Detection by Multi-Feature Classification\" -15 December 2023, DOI-r 10.1109/ACCESS.2023.3339621 [3] Matthew Carter and Michail Tsikerdekis \"Approaches for Fake Content Detection: Strengths and Weaknesses to Adversarial Attacks\" [4] Kuai Xu, Feng Wang, Haiyan Wang, and Bo Yang \"Detecting Fake News Over Online Social Media via Domain Reputations and Content Understanding\" -February 2020,DOI-10.26599/TST.2018.9010139

Copyright

Copyright © 2026 Chathrapathi P, Ahmed Aabith A, Yuvaraj K, Rasik Fareeth P, Vasu G. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.