Adapting existing natural language processing resources for cardiovascular risk factors identification in clinical notes