This paper presents a solution to extend the IEEE LOM standard with ontology-based semantic annotations for efficient use of learning objects outside Learning Management Systems. The data model corresponding to this approach is first presented. The proposed indexing technique for this model development in order to acquire a better annotation of learning resources is further presented. This technique extends and combines two consecrated alternative methods for structure-based indexing of textual resources: the mathematical approach of the latent semantic indexing and the linguistic-oriented WordNet-based text processing. Thus, the reason behind the good results provided by the first method becomes more transparent due to the linguistic controlled choices proposed by the second method. The paper results are important in the context of adopting semantic web technologies in the e-learning field, but also as a progress in the area of ontology-based indexing of textual resources.