|
|
 |
| |
|
|
|
Automated Child Detection on Smartphones through Behavioral Analysis Techniques |
|
|
|
PP: 235-251 |
|
|
doi:10.18576/amis/200116
|
|
|
|
Author(s) |
|
|
|
Asmaa M. Elsify,
Alaa Elnashar,
Ahmed Hamdy,
Ahmed Mahfouz,
|
|
|
|
Abstract |
|
|
| The use of internet-connected devices like smartphones and tablets has become integral to daily life, raising significant privacy and safety concerns for young users. Automatically distinguishing child users can empower devices to filter inappropriate content and provide a safer online environment. In this paper, we introduce kidDetect, a novel framework that detects children on smartphones by analyzing their distinctive interaction behavior, including touch dynamics and device holding patterns. We conducted a study with 198 participants, collecting touch and sensor data non-invasively as they used various applications. Our findings confirm that children exhibit statistically unique behaviors. By leveraging these patterns, we developed classifiers that achieve strong performance in child detection, with an AUC of 0.93 for swipe gestures and 0.90 for tap gestures.
|
|
|
|
|
|
|
|
