In this paper, we propose to investigate the incorporation of reputation mechanism in an e-learning environment in order to generate personalized recommendations. The proposed architecture, named e-RecRep, aims to allow the recommendation of LOs in an e-learning environment, where the reputation of users who recommend these LOs is considered. With the adoption of e-RecRep, the student receives suggestions - from the system and from other users - of learning objects that relate to the studied content, encouraging the student to complement his learning. Preliminary results allow us to conclude that suggestions from a person who presents a good reputation to a group make the recommended information more relevant, improving not only the credibility of the information but also its robustness, diversity and surprise (serendipity). The article will be structured as follows. The main concepts and related work will be presented initially. The RecRep architecture and its main features are then presented as well as the description of their behavior through real usage scenarios. Finally the results of the first experiments involving its use and the possibilities of continuity of the work are discussed.