KeyRecs: A keystroke dynamics and typing pattern recognition dataset.

Area of Science:

• Biometrics and Human-Computer Interaction
• Cybersecurity and Authentication Technologies

Background:

• Keystroke dynamics offer a passive and non-intrusive method for continuous user identity verification.
• Intelligent authentication systems can benefit from analyzing typing patterns to enhance security.
• Existing datasets may lack comprehensive typing behavior and demographic data for robust model training.

Purpose of the Study:

• To introduce and describe the KeyRecs dataset, a novel resource for keystroke dynamics research.
• To provide a dataset that supports the development of advanced biometric authentication systems.
• To facilitate research in user recognition and unauthorized access prevention using typing behavior.

Main Methods:

• Collected keystroke data (inter-key latencies) from 99 participants performing fixed-text and free-text typing exercises.
• Recorded digraph latencies, measuring time between key press and release events.
• Gathered demographic information including age, gender, handedness, and nationality for each participant.

Main Results:

• The KeyRecs dataset comprises diverse typing samples and associated demographic data.
• The data captures fine-grained temporal information of user interactions with a keyboard.
• The dataset is suitable for training and evaluating machine learning models for user authentication.

Conclusions:

• Keystroke dynamics, as captured in the KeyRecs dataset, are valuable for intelligent authentication.
• The KeyRecs dataset can significantly improve the accuracy of authorized user recognition.
• Leveraging this dataset aids in developing more effective biometric authentication software to prevent unauthorized access.