The Design of the TERENCE Adaptive Learning System

Nowadays, circa 10% of 7-11 olds turn out to be poor comprehenders: they demonstrate text comprehension difficulties, related to inference-making, despite proficiency in low-level cognitive skills like word reading. The use of more intelligent ALSs to customtailor reading interventions in the form of smart games for poor comprehenders has tremendous potential. The TERENCE ALS embodies that potential. This paper presents the design choices of the TERENCE ALS. First it specifies the preliminary user requirements of the system, and the system's intended usage through selected usage scenarios. The paper continues describing the system's learning material, made of stories and games, according to the preliminary users’ requirement specifications. The learning material and stories are structured as in the presented domain model, part of the conceptual model of the TERENCE architecture. Finally, the paper explains the architecture of the TERENCE system, focusing on the adaptive engine and intelligent backbone.