In this paper, we present a specific ontologies based on the extraction of semantic information in text documents. Document here concerns the context of the valorization of Tunisian patrimony. As approach, we propose to represent semantic properties in document contents from heterogeneous information (multimedia) concerning by the patrimony objects. For indexing and information retrieval (IR), we develop processes based on the noun phrase (NP) properties and their semantic representations. These processes use natural language processing (NLP) to take into account the NP syntactic and semantic structures. In view of this study, the specific ontology designed has the encapsulation principle to capitalize the concept and knowledge as NP and its semantic relations.