Although widely used, the SCORM metadata model for content aggregation is difficult to be used by educators, content developers and instructional designers. Particularly, the identification of contents related with each other, in large repositories, and their aggregation using metadata as defined in SCORM, has been demanding efforts of computer science researchers in pursuit of the automation of this process. Previous approaches have extended or altered the metadata defined by SCORM standard. In this paper, we present experimental results on our proposed methodology which employs ontologies, automatic annotation of metadata, information retrieval and text mining to recommend and aggregate related content, using the relation metadata category as defined by SCORM. We developed a computer system prototype which applies the proposed methodology on a sample of learning objects generating results to evaluate its efficacy. The results demonstrate that the proposed method is feasible and effective to produce the expected results.