Keystroke100 benchmark dataset is a dataset for research on pressure-sensitive keystroke dynamics and typing biometrics. This dataset contains keystroke/typing patterns of 100 users typing the password "try4-mbs".
A program was developed to collect keystroke latency and keystroke pressure from a total of one hundred computer users. The participants were requested to familiarise themselves with the password "try4-mbs" prior to the actual data collection session. The use of a common password among all participants is to ensure the comparability of typing patterns of the same password from different individuals. A total of ten timing samples and ten pressure samples were collected from each participant.

There are two files in each user's folder, i.e. latency.txt and pressure.txt.

Each row of the matrix in latency.txt corresponds to a sample whilst each column corresponds to a latency (time gap between two keystrokes). Keystroke latency data samples were captured at an accuracy of milliseconds (ms).

Each column in the pressure.txt corresponds a sample, i.e. a time-series of pressure (measured in volt) exerted on a key.

The dataset is intended for research purposes only and as such cannot be used commercially. In addition, reference must be made to the following publications when this dataset is used in any academic and research reports.


  1. Keystroke Patterns Classification using the ARTMAP-FD Neural Network
    C. C. Loy, W. K. Lai, and C. P. Lim
    International Conference on Intelligent Information Hiding and Multimedia Signal Processing, Taiwan, 2007 (IIHMSP 2007)
  2. Pressure-based Typing Biometrics User Authentication Using The Fuzzy ARTMAP Neural Network
    C. C. Loy, C. P. Lim, and W. K. Lai
    International Conference on Neural Information Processing, Taiwan, 2005 (ICONIP 2005)