The Domain Model of an Adaptive Learning System for Poor Comprehenders

Abstract

Comprehension of stories involves the decoding of the text, the activation of prior knowledge and, ultimately, the construction of a coherent mental representation of what the story is about. Nowadays, more and more 8-10 year-old children turn out to be poor comprehenders: they demonstrate text comprehension diculties, despite pro ciency in decoding and well developed low-level cognitive skills, like vocabulary knowledge. A few adaptive learning systems promote reading interventions, but not for poor comprehenders. The TERENCE EU project aims at building a learning system that adapts its learning material to the speci c needs of poor comprehenders. For creating the conceptual model of the TERENCE system, we rst need to understand (1) who the end users are, and (2) what the learning material is. This thesis rst analyses and speci es the user requirements of poor comprehenders, both hearing and deaf, thereby laying the groundwork for building the ontologies of the conceptual model of TERENCE. The main learning material of TERENCE will consist in illustrated stories, and interactive games for reasoning about the stories. The domain sub-model of the TERENCE conceptual mode structures this learning material. It is thus composed of two main ontologies: one for stories, the other for reasoning games. The main objective of this thesis is to start building the domain ontologies of the TERENCE adaptive learning system. To the best of our knowledge, our domain ontologies are the rst pertaining to stories (story ontology) and reading interventions (game ontology) speci c for poor comprehenders. Our engineering of the domain ontologies meant analysing the following material: (1) concept schemes and concepts pertaining to text diculty, mainly for the story ontology, and (2) several taxonomies of reading comprehension and reading interventions, mainly for the game ontology. The material and the resulting ontologies were assessed under the guidance of the respective domain experts. Specifying the context of use and the requirements of poor comprehenders also help us shape the student model of the TERENCE system. This model is again realised as an ontology. This thesis concludes presenting the purposes, the expected format and sources (ontologies and not) for the student model, as well as its relations with the domain ontologies and the user requirements speci ed in this thesis.