Learners automated evaluation with the ODALA approach

We present in this paper ODALA approach (Ontology-Driven Auto-evaluation for e-Learning Approach) for an automated evaluation of the learners state of knowledge. The context considered is Computer Based Human Learning Environnement (CBHLE) in a self-learning by doing mode. This approach, that we put in work and test in the setting of a self-learning system for an algorithmic language, is founded on the representation of the teaching domain as domain ontology on the one hand and on errors classification and detection on the other hand. The evaluation process that we recommend is structured in four stages, starting from the learner solution form analysis (that consists at this step of our research to a lexico-syntactic analysis) and finish with the update of the learner model, while passing by a semantic analysis and a marking process. After a brief introduction of the main aspects raising of the CBHLE domain to which we refer here, we develop our approach of the learners evaluation problem, while especially insisting on its independence of the teaching domain and on the possibility to take in account answers to open questions, freely built by the learner. We also present, at the end, the results of the algorithmic self-learning system development, where the main stages of our evaluation approach are implemented.