Data mining methods application in reflexive adaptation realization in e-learning systems

• Main Authors: A. S. Bozhday, Y. I. Evseeva, A. A. Gudkov
• Format: Article
• Language: English
• Published: Plekhanov Russian University of Economics 2017-09-01
• Series: Otkrytoe Obrazovanie (Moskva)
• Subjects: adaptation; virtual educational environment; e-learning; information systems
• Online Access: https://openedu.rea.ru/jour/article/view/423
• ISSN: 1818-4243; 2079-5939

In recent years, e-learning technologies are rapidly gaining momentum in their evolution. In this regard, issues related to improving the quality of software for virtual educational systems are becoming topical: increasing the period of exploitation of programs, increasing their reliability and flexibility. The above characteristics directly depend on the ability of the software system to adapt to changes in the domain, environment and user characteristics. In some cases, this ability is reduced to the timely optimization of the program’s own interfaces and data structure. At present, several approaches to creating mechanisms for self-optimization of software systems are known, but all of them have an insufficient degree of formalization and, as a consequence, weak universality. The purpose of this work is to develop the basics of the technology of self-optimization of software systems in the structure of e-learning. The proposed technology is based on the formulated and formalized principle of reflexive adaptation of software, applicable to a wide class of software systems and based on the discovery of new knowledge in the behavioral products of the system.To solve this problem, methods of data mining were applied. Data mining allows finding regularities in the functioning of software systems, which may not be obvious at the stage of their development. Finding such regularities and their subsequent analysis will make it possible to reorganize the structure of the system in a more optimal way and without human intervention, which will prolong the life cycle of the software and reduce the costs of its maintenance. Achieving this effect is important for e-learning systems, since they are quite expensive.The main results of the work include: the proposed classification of software adaptation mechanisms, taking into account the latest trends in the IT field in general and in the field of e-learning in particular; Formulation and formalization of the principle of reflexive adaptation in software systems applicable to a wide class of applied programs; The development of a universal architectural template of the software system, oriented to restructuring in the process of operation; Algorithm for self-optimization of the user interface of the software system based on methods of data mining.The development of the theoretical basis for the automatic reorganization of e-learning software will increase the flexibility of the virtual educational environment and increase the period of its exploitation. Unlike existing analogues, the methods proposed in the article are universal and applicable to a wide class of applied programs. This is relevant for e-learning systems, because their may have a different types and purposes (for example, virtual simulators and information library software may be components of one system).