This thesis addresses improvements in recommendation systems for student activities through the use of automatically generated concept maps. It focuses on the development and analysis of a concept map generation system as a basis for a recommendation system for student activities.
The initial phase of the research is dedicated to optimizing the learning process, the role of recommendation systems, and the utility of concept maps. The aim of the thesis is to explore how automatically generated concept maps can achieve results comparable to manually generated ones. This was determined by evaluating these concept maps.
The research includes the use of natural language processing algorithms to generate keywords and create concept maps from textual data. Test data for both types of maps, manually and automatically generated versions, were analyzed. Statistical significance tests, based on student feedback, were used to compare the quality of manually and automatically generated maps.
The research results showed that automatically generated concept maps can successfully replace manually created versions, confirming their usefulness in recommendation systems. Survey results among students indicated that automatically generated maps align well with their needs, with no noticeable differences between automatically and manually generated maps. The conclusion of the research found that further algorithm development is necessary, particularly for filtering irrelevant words, along with expanding the research with a larger dataset and a more extensive group of evaluators.
|
