At the last decades Learning Objects Repositories have had an ex- plosive growth given the state policies that favored their development in the context of open access. However, this growth was not ordered, resulting in the use of different metadata standards, different policies for filling them, different interpretations, etc. As a consequence, the interoperability between Learning Object Repositories was weakened, mainly at the semantic level. In this article we propose the implementation of a shared vocabulary represented by an on- tology, which is validated based on queries made to it.