Network mining has been traditionally used in different application domains in order to enhance the Content that Web users are accessing. Likewise, Website administrators are interested in finding new approaches to improve their Web site content according to their users' preferences. Furthermore in Semantic Web has been considered as an alternative to represent Web content in a way which can be used by intelligent techniques to provide the organization, meaning, and definition of Web content. In This work, we define the Website Key Object Extraction problem, whose solution is based on a Semantic Network mining approach to extract from a given Website core ontology, new relations between object according to their Web user interests. This methodology was applied to areal Website, whose results showed that the automatic extraction of Key Objects is highly competitive against traditional surveys applied to Web users.