Diagnosing Knowledge using Learning Activity Traces Generated by Various Problems Solving Modalities
Abstract
Learners work increasingly with a large panel of learning environments that involve them in various activities thanks to a number of tools. These tools generate activity traces, which must then be taken into consideration and combined so as to establish the most accurate diagnosis about the learner’s activity. This paper presents a diagnosis model, called DiagElec. This model considers traces generated by various independent tools. DiagElec integrates a notion of uncertainty in the diagnoses that it generates thanks to the notion of ‘degree of belief’ which is defined by the rules of the diagnosis. To evaluate our model, we carried out a two-phase experiment, first with learners and then with teachers. From the corpus of diagnoses compiled by the teachers, we look for the emergence of a model of human behaviour in order to ascertain whether or not it is necessary to readjust the degree of belief defined by the rules of the diagnosis.