Ontology alignment is grasping an increasing attention topic. Indeed, it is a palliating issue to the heterogeneity problem that arises among the semantic web data. In this respect, a steady effort within the knowledge management community, and ontology alignment in particular, is carried out to meet the thriving raised challenges. Indeed, it is of paramount importance to overcome the natural language barrier as well as the produced alignments reuse and exploitation. As far as the challenging cited issues are both intrinsically related to the basic alignment problem, our contribution aims to provide a framework encompassing two alignment systems. The first one is dedicated to direct cross-lingual ontology alignment. The second one treats the indirect ontology alignment issue. The underlying idea of our approach is that the linguistic heterogeneity and its after-effect are reduced thanks to well-defined techniques use and harmonization. Those techniques endow our methods with more efficiency and accuracy in the candidate mappings detection phase. In the same way, we introduce an alignment composition system in order to valorize existing ones by producing indirectly computed alignments. We evaluated our contribution in various tracks and obtained a high-quality encouraging results.