Security Threat Exploration on Smart Living Style based on Twitter Data

Abstract

The Internet of Things (IoT) has revolutionized individuals’ homes with smart devices, but it has also brought security worries due to the huge amounts of data they generate. This study aims to uncover common security problems, like malware, cyber-attacks, and data storage flaws, in such smart setups. To tackle these issues, this study suggests beefing up security measures and educating users about safe device practices. A new approach was followed in this study, using Convolutional Neural Networks (CNNs) instead of the traditional Natural Language Processing (NLP) methods. CNNs are great at understanding complex patterns in text, especially on platforms like Twitter where messages can be brief and unclear. By applying CNN to analyze Twitter data, specific entities linked to security issues could be pinpointed, giving a deeper insight into smart home security challenges. The findings showed that the employed CNN model was exceptionally efficient at sorting out tweets regarding security problems in smart homes. It achieved an accuracy of around 87%, precision of 76.78%, recall of 82.49%, and F1-score of 84.87% surpassing the other methods it was compared with. These findings underscore the CNN model's effectiveness in accurately classifying security-related tweets in diverse topics within smart living environments.