This disambiguation approach use semantic similarity techniques to rank candidate entities solely according to their lexical similarity and/or empirical co-occurrence with mentions.