Banca de DEFESA: MARCELLO RAMALHO DE MELLO

LEARNING IN OPEN AND MASSIVE ONLINE COURSES (MOOC):

 a proposal for automating the assessment by concept maps.

Assessment, MOOCs, Meaningful Learning.

Massive Open Online Courses (MOOCs) experienced significant growth in 2021, with 220 million students worldwide (excluding China) and offering thousands of courses and degrees. In Brazil, CAPES announced 100,000 free spots in online courses, totaling 300,000 spots since 2020. With the rise of online courses, there is a need for effective learning assessment as traditional methods have limitations. Online assessment requires rethinking the process and leveraging technology to provide personalized, automated, and diverse feedback while collecting data on learning outcomes. In this context, the evaluation through Concept Maps in MOOCs emerges as a promising approach to understanding how students relate concepts and automating the assessment process. The present study aims to propose and evaluate a model for automating the assessment of learning with Concept Maps in massive open online courses (MOOCs). The study was conducted in four stages: 1) identifying system attributes; 2) proposing the solution; 3) validating the mechanics of operation; 4) validating the results provided by the solution. The presented solution is a computer algorithm called ACME, coded in Python, which extracts data from Concept Maps and provides a numerical summary through comparison with a base text, enabling massive and automated evaluation. The mechanics of the algorithm were validated with a base text and a concept map, allowing for the discovery of potential variables for analysis and assessment of learning. The results provided by the algorithm were validated by comparing them with an article that provided analysis of concept maps created by students, including variables such as the quantity of concepts, correct concepts, number of propositions, verb-linked terms, Jaccard similarity, cosine similarity, and spaCy similarity. Graphical analysis was also obtained. The study obtained promising results for the idea of evaluating student learning qualitatively and on a massive scale through Concept Maps. It is up to the instructor, who decides to use the proposed solution, to build their assessment models and obtain both quantitative and qualitative parameters to identify and evaluate mechanical and meaningful learning.