In view of a low accuracy rate and recall rate of named entity recognition in the automotive industry, a new method of named entity recognition based on ontology has thus been proposed. By extending the existing thesauri, and constructing an automobile domain ontology, the ontology features are to be extracted from the corpus, and a named entity recognition based on a CRFs model can be achieved. The experimental results show that the ontology features can effectively identify the vehicle attribute entities, with its accuracy rate as high as 75.60%, a recall rate as high as 66.12% and a F-value as high as 70.54% respectively.