Ubiquitous Learning Environments (ULEs) - or Technology Rich Classrooms - are created to combine real and virtual learning environments in order to produce richer study artifacts to support both instructors and students in the teaching/learning process. However, they produce a huge volume of Learning Objects (LOs), often unstructured, which makes retrieving and presenting them a challenge, especially when presented in an individualized manner. In this work, we explore the concept of Adaptive Educational Hypermedia (AEH) together with ULEs for creating and providing personalized LOs to students. This research proposal aims at designing a component-based architecture capable of structuring automatically captured LOs in classrooms and their metadata through a collaborative approach, and offering content personalization features in ULEs, considering social and collaborative interactions, devices' contextual constraints, and personal preferences and traits. Such architecture is being integrated into a real multimedia capture platform currently used as a complementary tool by instructors of the Faculty of Computing at Federal University of Uberlândia, Brazil.