Keyboard Handwriting as a Factor of User Authentication

Currently, information technologies are used in all spheres of our life. Huge amounts of personal and corporate information are stored digitally, the disclosure of which can bring serious problems to an individual or a company, may entail reputational losses or financial damage. In this regard, there is a problem of protecting this information, ensuring its security. Currently, one of the most common methods of information protection is means of identification and authentication. According to modern information security standards, information about the dynamics of the user’s work with the keyboard can be used as authentication data. The development of an continual authentication system based on the dynamics of the user’s work with the keyboard is divided into 2 subtasks: the formation of a dataset from raw data by identifying features and training the classifier on the prepared data. At the stage of forming suitable data sets, the source data is divided into windows, and a set of statistical features for analysis is allocated in each window. At the stage of using the classifier, the optimal parameters of the classifier for training are selected. The resulting model is able to accurately determine the legitimacy of the user by a set of characteristics of the dynamics of his work with the keyboard. Moreover, it provides opportunities for further research on the possibility of building an continual authentication system based on the analysis of keyboard handwriting.