Intelligent Tutoring Systems (ITSs) are educational software containing some kind of intelligent component to imitate how a human teacher would behave when teaching. Adaptive Educational Hypermedia systems (AEHSs) are inspired in ITSs to adapt the content and navigation in the course to each student’s model. All the same, the assessment of these systems is usually focused on the so-called objective testing (Multiple-Choice Questions, fill-in-the-blank items, etc.), which is commonly agreed that fails to identify many students’ deep underlying misconceptions. Hence, automatic assessment of free-text answers is a field that has attracted much attention in the last decades. About twenty different systems are being used both in academic and commercial environments underpinned by several Natural Language Processing (NLP) techniques. Nevertheless, none of these systems keeps any kind of student model.
It has been highlighted the importance of keeping a student model that serves to identify the lack of previous knowledge and to let the system decide which questions are the most suitable. Therefore, in this work, ideas from the AEH, NLP and CAA fields are combined to build a bridge between what teachers try to transmit and what students actually understand. It is achieved by a new procedure able to automatically generate the students’ conceptual models from their answers to a free-text Adaptive Computer Assisted Assessment (ACAA), which is the evolution of free-text CAA systems but incorporating the use of a student model to adapt the assessment. Using this approach, it is possible to generate each particular student’s conceptual model and the whole class conceptual model.
In this way, teachers can visualize at any time the degree of assimilation of
the concepts exposed in the lessons and, to discern their students’ faulty or
incomplete knowledge in order to organize more efficiently the agenda of their
courses. Furthermore, by using the free-text ACAA system, students can get
instant feedback from their answers (score, processed answer and correct answers
provided by the teachers) and, by looking at their conceptual models, they
can organize their study, i.e. review which concepts have already assimilated
and, which ones are still missing or are wrongly connected to other concepts.
The procedure has been implemented in the Will tools that consist of: Willow, a
free-text ACAA system; Willed, an authoring tool; Willoc, a configuration tool;
and COMOV, a conceptual model viewer in which several representations have been
used to show the conceptual models (a concept map, a conceptual diagram, a
table, a bar chart and a textual summary).