Once drug names in the input text are identified, Peters et al. [2011] further compare candidate strings in the RxNorm dataset with the tokenized input text to determine its terminology concept, with a string similarity measure based on the Jaccard coefficient.