A behaviour biometrics dataset for user identification and authentication.

Area of Science:

• Cybersecurity and Machine Learning
• Behavioral Biometrics

Background:

• Online transactions generate sensitive personal data, making users vulnerable to identity fraud.
• Existing security measures are insufficient against sophisticated cyber threats.
• Behavioral biometrics, such as keystroke and mouse dynamics, offer a unique and difficult-to-replicate user authentication method.

Purpose of the Study:

• To introduce a novel dataset of behavioral biometrics for cybersecurity research.
• To facilitate the development of machine learning models for detecting fraudulent online payment activities.
• To enhance user authentication in e-commerce through continuous, passive monitoring.

Main Methods:

• Collected keystroke and mouse dynamics data from 88 individuals.
• Created a dataset with 1760 instances, labeled as legitimate or illegitimate card owner behavior.
• Utilized machine learning techniques for anomaly detection in user behavior patterns.

Main Results:

• The presented dataset enables the training of machine learning models to differentiate between genuine and fraudulent user behavior.
• Behavioral biometrics demonstrate potential as a robust layer of security in online transactions.
• The dataset is publicly available for academic research in cybersecurity.

Conclusions:

• Behavioral biometrics, specifically keystroke and mouse dynamics, present a promising approach to combatting online identity fraud.
• Machine learning analysis of these biometrics can significantly improve the accuracy of user authentication.
• The 'Behaviour Biometrics Dataset' supports ongoing research in creating more secure e-commerce environments.