IRMA-International.org: Creator of Knowledge
Information Resources Management Association
Advancing the Concepts & Practices of Information Resources Management in Modern Organizations

Automating Child Protection: The Future of AI, ML, and Digital Safety

Automating Child Protection: The Future of AI, ML, and Digital Safety
View Sample PDF
Author(s): Hitesh Rawat (University of Extremadura, Spain), Anjali Rawat (University of Extremadura, Spain), Prathamesh Muzumdar (Mangalayatan University, India), Shyam Gehlot (Shri Vaishnav Vidyapeeth Vishwavidyalaya, India)and Chandrapal Singh Dangi (Manipal University Jaipur, India)
Copyright: 2026
Pages: 24
Source title: Integrating Parental Consent and Child Engagement With Digital Protection Rules
Source Author(s)/Editor(s): Romil Rawat (LabGeoInf-Research LABoratory in GEOmatics and INFormation systems, National Research Council in Italy, Rome, Italy), Sanjaya Kumar Sarangi (Utkal University, India), A. Samson Arun Raj (Karunya Institute of Technology and Sciences, India), Janet Olivia Richmond (Karunya Institute of Technology and Sciences, India)and Purvee Bhardwaj (Rabindranath Tagore University, India)
DOI: 10.4018/979-8-3373-2716-7.ch002

Purchase

View Automating Child Protection: The Future of AI, ML, and Digital Safety on the publisher's website for pricing and purchasing information.

Abstract

With increasing digital exposure, safeguarding children online is more important than ever. This research introduces an AI-driven framework, ChildGuard-CNN, designed to detect and prevent harmful online content targeting children. The system combines Word2Vec embeddings with a Convolutional Neural Network (CNN) to analyze and classify digital content. Training and validation were conducted using two public datasets: the 2023 Cyberbullying Detection Dataset (45,000+ social media posts) and the YouTube Open Dataset for Content Moderation (60,000 instances of metadata flagged for safety concerns). Word2Vec transforms text into semantic vectors, which are processed by the CNN for feature extraction and classification. The model achieved 95.37% accuracy, with 94.89% precision, 95.78% recall, and a 95.33% F1-score. It also supports real-time detection with an average processing time of 1.3 seconds. Future improvements aim to enhance multilingual support and broaden platform compatibility for greater reach and effectiveness.

Related Content

Ayah Abdulraheem, Mohammad Arafah. © 2026. 30 pages.
Hitesh Rawat, Anjali Rawat, Prathamesh Muzumdar, Shyam Gehlot, Chandrapal Singh Dangi. © 2026. 24 pages.
Andi Asrifan. © 2026. 38 pages.
Romil Rawat, Kuldeep Singh, A. Samson Arun Raj, Prathamesh Muzumdar, Sanjaya Kumar Sarangi, Hitesh Rawat. © 2026. 24 pages.
Nandini Bansod, Akash Mishra, Dinesh Baban Kamble, Ashutosh Shukla. © 2026. 20 pages.
Neeta Vaibhav Mhavan, Priya R. Singh. © 2026. 26 pages.
Vinod Mahor, Jaytrilok Choudhary, Dhirendra Pratap Singh. © 2026. 28 pages.
Body Bottom