Knowledge management has increasingly become a big challenge, especially in large organizations. The known vision of the Semantic Web is to offer intelligent services made possible by facilitative machine understanding of web content, especially by using ontologies. This work proposes an ontology-based approach for knowledge management, which stems from model based on a network of ontologies capable of representing knowledge from structural and syntactically heterogeneous resources, and of building an organizational knowledge graph based on a network of ontologies. This approach is composed of a Profile Ontology that defines the shared vocabulary of a specific organization and a set of resource representation ontologies created automatically via a knowledge extraction process that deals with structured, semi-structured and unstructured resource formats. This methodology has been successfully tested by creating an organizational memory system for a specific organization by defining its Profile Ontology and building a network of ontologies created from its varied resources.