Study of the possibilities to use keyboard handwriting for the tasks of identifying students in e-learning systems

Abstract

When using distance education systems, the problem of information security of the educational process arises, which, in addition to external ones, also implies internal threats. One of these threats can be a legitimate user who paid a fraudster to take tests and give visibility to educational activities under his own name. The use of traditional identification methods has two significant drawbacks: firstly, the ambiguity of the identified user, because the identification of the user occurs by the entered pair login-password; secondly, the inability to detect the substitution of an identified user in the process of working with the system. These disadvantages are eliminated by using biometric methods of covert and continuous monitoring.

In the first part of the work the different types of control knowledge tests are analyzed. Taking into account the specifics of the use of covert keyboard monitoring algorithms, the following is proposed: 1) to use tests that do not contain answers; 2) use tests after each learning activities in order to form a user’s biometric vector; 3) use tests with numerical answers in order to minimize the analyzed keystroke digraphs.

An algorithm for user’s profile formation and its identification is proposed in the second part of the work. Its combine qualitative (distribution of the frequencies of using numeric keys groups, comma-separated keys, “plus” and “minus” keys on the main and additional keyboard units) and quantitative (analysis of statistical properties of keystroke digraphs) approaches. The experimentally obtained estimates of the identification accuracy of the proposed algorithm: FAR=4.64% and FRR=6.25%.