Uma técnica formal para avaliação automática e geração de feedback no contexto da educação digital

Abstract

Modern schools try to integrate the advantages of information technology in their pedagogical means. However, these attempts are usually limited to just attaching to the physical environment, hardware artifacts so that students and teachers can use them. When seeking to incorporate technological culture into education, they also seek to incorporate new forms of teaching and learning. Therefore, it is not enough to use alternative resources and state-of-the-art technologies unless strategic capacities are developed on how and when to use such resources in the teaching-learning process. The task of assessing and classifying student performance, in addition to providing information based on these assessments, is one of the points in which integration with technology seeks to act. This automation brings productivity and time management gains for both teachers and students. Automatic assessment is an algorithmic technique used to prove/reflect that a student’s response meets criteria established by the teacher. When a response is classified as "incorrect", the assessment follows the feedback process, which consists of providing information about the student’s performance, aiming to reorient or stimulate more appropriate future behavior through recommendations. In this sense, this work presents a methodology that seeks to automate the task of evaluating a student’s performance and provide feedback to students and teachers in the digital education scenario. Unlike other works with similar objectives, this proposal seeks to use a formal method, based on Petri net, as a means of ensuring the consolidation and systematization of information to be evaluated. The automatic evaluation is presented as a function of the formal modeling of student behavior when responding to an evaluation activity. The feedback is based on the verification of properties and components of this model. A systematic review of the literature was performed where eleven methods were identified, ten of them related to the evaluation and six to the feedback. However, there are few works that use formal methods to automate the evaluation and generate feedback. Case studies of the application of the proposed method were conducted, where situations of students are demonstrated from two perspectives: the pedagogical, which relates the metrics with the final result of the evaluation of the formal model of student behavior; and the technique, which shows how the criteria that the teacher selects can be met from the elements and/or properties of the model to generate feedback on student performance in the evaluation. The results show that the proposed methodology is promising because it can efectively prove the correctness of the assessment and, additionally, provide feedback to students and teachers when they present some flaw in performance.