A Network Matriculation Management System Based On Keystroke Dynamics Authentication

In the process of matriculation, it is normally required to use the network-based matriculation system developed by Moe (Ministry of Education). The system helps the universities to connect with all the provincial data servers, share the candidates information, examine electro-dossiers, and complete the matriculation task. Afterwards, each university should start the subsequent data classification and processing, which is a huge job but the existing system developed by MoE cannot handle it. It is therefore the aim of this thesis to develop an extension system for the university to deal with these tasks efficiently and timely, which can be linked to the existing system by MoE.The system consists of mainly the following modules, classification and backup, data pre-processing, printing and delivering the notification to the accepted students, classified data statistics, data collection and maintenance, students grouping, name list making, and so on. The developed system can process effectively the provincial data and local data in a unified manner, and make the student data be matched coherently with the internal university management information system. In this way, the system enhances significantly the automation degree as well as the working efficiency.In order to better protect student data and to strengthen system security, an identity authentication module based on keystroke dynamics is developed. The module adds an additional biometrics protection to the existing password protection since users can login the system safely even if the password is disclosed. When a user login, the system collects his keystroke characteristics, then process the collected data by based on fuzzy-logic and statistics algorithms. The user's keystroke characteristics can then be figured out, and the user identity can be confirmed. In the fuzzy-logic algorithm, by first using the data statistical taxonomy based on golden section, the collected data are classified into different classes according to the speed of the keystroke and the pre-determined threshold, therefore resulting better identity authentication.