Abstract: This paper proposed an approach called WOSR for the semantic retrieval of domain resources that have been previously annotated with ontology concepts. Firstly WOSR developed domain ontology and associated weights with concepts through uniform probability distribution. Secondly the similarity between concepts was computed by using their weights. Finally calculated semantic similarities between a user request and the domain resources, and the answer was returned in the form of a ranked list ordered by semantic similarity. Experiments show that WOSR yields a better performance than other classical methods.

Key words: Semantic retrieval, Domain ontology, Similarity