The aim of this project is to study and develop an adaptive planning system for the automated and intelligent composition of educational resources in the semantic web domain. The purpose of the system is to construct personalized curricula for distant e-learning courses, based on the knowledge state and the educational goals of the learner. The first goal of the project is to augment the HAPRC planning system, which combines techniques from automated planning, machine learning and knowledge-based systems. The improvements in the planning system focus on: a) efficiency of the planning algorithm in terms of running time and quality of plans, b) ability to handle meta-data of educational resources expressed in XML and RDF and c) representation and solving of curricula synthesis problems.