KeyRecs is a keystroke dynamics dataset that can be used to train, validate, and test machine learning models for anomaly detection and robust typing pattern recognition, as well as the clustering and classification of users that present a similar behavior. It contains fixed-text and free-text samples of user typing behavior, obtained in a study with 100 participants of 20 different nationalities performing password retype and transcription exercises. The samples consist of inter-key latencies computed by measuring the time between each key press and release during an exercise, following a digraph model. Additionally, the participants were also asked to provide their demographic information regarding age, gender, handedness, and nationality. KeyRecs can be valuable to enhance the recognition of authorized users and prevent illegal logins in biometric authentication software, and can be combined with additional data recordings to create more extensive datasets and improve the generalization of machine learning models. If you use this dataset, please cite the primary data article: https://doi.org/10.1016/j.dib.2023.109509

This work was partially supported by the Norte Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, through the European Regional Development Fund (ERDF), within project “Cybers SeC IP” (NORTE-01-0145-FEDER-000044). This work has also received funding from the projects For-Pharmacy (P2020-COMPETE-FEDER nr 070053) and UIDB/00760/2020.