To overcome the shortcomings of traditional search engines with the limitation to the retrieval of lists of potentially relevant document, an ontology-based intelligent search engine, called CRAB, aims to deploy natural language tools to automatically extract knowledge from web documents and effectively manage RDF triples in a given Knowledge Base by an OWL DL reasoner, which is used to check the consistency for KB, inferring implicit, duplicate and inconsistent knowledge. In this paper we propose context-profile based approaches for automatic instance identification applied in the CRAB System. In our experiments we populate the CRAB Ontology with new instances identified by presented approaches. Both theory and experimental results have shown that our methods are inspiring and efficient.