UL-Keystroke: A Web-Based Keystroke Dynamics Dataset

Keystroke dynamics-based authentication systems identify individuals by analyzing their keystroke patterns when interacting with input devices such as a computer keyboard. Within the fields of Statistics and Machine Learning, several research studies have applied different techniques for recognizing keystroke patterns. This work proposes the creation of a dataset and a methodology that would allow users to capture typing patterns from students at a university in Lima, Peru, using a cloud environment and their personal devices. The cloud architecture used for the implementation and deployment of the web tool will be explained in detail. The result of this work is a dataset containing participant information, records of their keystroke patterns, and additional metadata from their web browsers, which could be used to enrich further studies. Moreover, in addition to the captured raw data, some keystroke dynamics features were generated and made available along with the dataset to facilitate the development of classification models. The dataset and methodology presented in this article can be used by other researchers to enhance existing keystroke dynamics recognition systems.