The semantic web (SW) realization leads researchers and IT professionals to design and develop many knowledge based (KB) applications. Ontologies and other structured vocabularies are used as their knowledge organization system (KOS). Research scientists and practitioners use ontologies to annotate data for facilitating data sharing and integration. The Linked Open Data initiative has opened new perspectives for integrating heterogeneous data on the web. The increasing availability of KOS on the web as well as ad hoc classifications built for specific purposes and not yet present in the infrastructure of the Semantic Web raises the question of their identification and reuse. In this perspective, we have designed and implemented ServO (Server of Ontologies), a tool for fast building distributed ontologies repository. The repository is then used to compute similarity between their entities. The ontologies, described in the RDF and OWL standard languages can be dynamically managed within the repository. ServO is available as a Java API and is based on IR techniques. A web client prototype has been implemented to showcase some functionalities of the tool.