IDENTIFICAÇÃO DE SUBSUNÇORES A PARTIR DE ANÁLISE QUALITATIVA BASEADA EM LÓGICA DIFUSA

Abstract

Automated assessment in Education has intensively used the Item Response Theory (IRT) model in the last decades. The IRT is a quantitative statistical model for assessing students' effective knowledge. In it, the students' scores are compared with scores defined by pre-testing the items of the assessment instrument. The Meaningful Learning Theory (TAS) advocates four stages in the teaching process: an initial stage, related to the survey of prior knowledge (subsumers) of each student, an advanced organization stage, which seeks to organize the subsumers raised, a stage related to the progressive differentiation of the concepts that one seeks to teach, and the integrative reconciliation stage. The subsumer survey stage is essential for the smooth running of the application of the TAS in teaching. However, this is a process for which quantitative assessments such as the IRT are not suitable, as the subsumption refers to the moment when the concepts are not yet well-grounded in the student's cognitive structure – they are, therefore, diffusely established. Fuzzy Logic becomes, then, a natural and an effective possibility for automated qualitative assessment of students since it can access concepts that have not yet been effectively stabilized in the cognitive structure of these students. Furthermore, fuzzy modeling is easily implemented computationally. This paper presents a general fuzzy modeling scheme for the problem of raising students’ subsumers to overcome the difficulties pointed out. This modeling has already been computationally implemented and is already in use, showing that the fuzzy qualitative approach to the problem is fully satisfactory for subsumer survey situations.