Análise automática da resposta do aluno em ambiente virtual

Abstract

Automatic analysis of student's answers to open questions has been rarely used in computer-mediated teaching, not only for theoretical restrictions to a methodology that is sometimes associated with the idea of programmed instruction, but also for practical reasons, due to its notorious implementational difficulties. This article describes an authoring system that was developed to offer the student an interactive learning environment, including answer analysis and the delivery of learning strategies. The system was developed according to three principles: (1) variable tolerance, in which more and less correct answers are accepted, (2) multiple segmentation, which allows for searching the most relevant elements in the student's written answer, and (3) diversified feedback, itself subdivided into generic feedback (simply tells the students if answer is correct or not), situated feedback (identifies student's specific needs), and strategic feedback (provides hints to help student construct meaning from text). The authoring system was piloted with teachers and students. The main conclusions was that it could contribute not only to improve distance learning but also classroom learning and teacher development.