Razvoj adaptivnog sistema za elektronsko učenje zasnovanog na nauci o podacima

Abstract

Učenje je okosnica savremenog društva, koje je pretrpelo globalne promene, u smislu digitalizacije i upotrebe IKT. Nameće se pitanje e-učenja, koje je aktuelizovano tekućom pandemijom. Kod e-učenja je potrebno osmisliti i sprovesti u delo jedan spoznajni proces usvajanja novih pojmova, savlađivanja umeća primene stečenih znanja u realnom okruženju i postavljanja temelja za dalje produbljivanje znanja iz obrađene oblasti. Uvek aktuelni problem je prilagođavanje procesa učenja potrebama učenika, tj. personalizacija učenja. Adaptivnost učenja je izazov koji je prenet i na polje e-učenja. Oblasti istraživanja disertacije odnose se na e-učenje i nauku o podacima. Predmet disertacije je razvoj i implementacija adaptivnog upravljačkog sistema za eučenje i odgovarajućih analitičkih modela nauke o podacima radi efikasnog donošenja optimalnih odluka. Ovo uključuje definisanje metodologije u realizaciji adaptivnog LMS, kao i sam njegov razvoj. Zadatak je složen i multidisciplinaran, jer uključuje oblasti poput sistema za upravljanje učenjem, nauke o podacima (Data Mining, skladištenje i obradu podataka, statističke metode, izveštavanje itd), internet tehnologija, ekspertskih sistema, kao i didaktiku i metodike. Nastava putem razvijenog i implementiranog adaptivnog modula e-učenja je sprovedena na Pedagoškom fakultetu Užice, Univerziteta u Kragujevcu. Rezultati sprovedenog eksperimenta su obrađeni statističkim metodama i primenom modela Data Mining i nauke o podacima. Razvijeni su i primenjeni statistički modeli za analizu podataka i specijalizovani Data Mining modeli za inteligentnu analizu podataka iz sistema za e-učenje, ekstrahovanje znanja i predviđanje. Na taj način je adaptivni sistem za e-učenje proširen modulom za Data Mining. Dobijen je bogat skup informacija koji je pokazao opravdanost primene adaptivnog LMS-a.

Learning is the backbone of modern society, which has undergone global changes, in terms of digitalization and the involvement of ICT. The issue of e-learning arises, which has become relevant due to the current pandemic. In e-learning, it is necessary to design and implement a cognitive process of acquiring new concepts, mastering skills of applying the acquired knowledge in a real environment and laying the foundations for further deepening of knowledge from the processed area. The always current problem is adaptation of the learning process to the needs of students, ie. personalization of learning. The adaptability of the learning is a challenge that has been transferred to the field of e-learning. The research areas of this doctoral dissertation relate to e-learning and data science. The subject of the dissertation is the development and implementation of an adaptive learning management system for e-learning and appropriate analytical models of data science for efficient decision making. This includes defining the methodology in the implementation of adaptive LMS, as well as its development. The task is complex and multidisciplinary, as it includes areas such as learning management systems, data science (Data Mining, data storage and processing, statistical methods, reporting, etc.), internet technologies, expert systems, as well as didactics and teaching methods. Teaching through the developed and implemented adaptive e-learning module was conducted at the Faculty of Education in Uzice, University of Kragujevac. The results of the conducted experiment were processed by statistical methods, Data Mining and data science methods. Statistical models for data analysis and specialized Data Mining models for intelligent data analysis from e-learning systems, knowledge extraction and prediction, have been developed and applied. In this way, the adaptive e-learning system has been extended with a module for Data Mining. A rich set of information was obtained, which showed the justification of the use of adaptive LMSs.