‘…detected ankyrin G,   which was…’	get raw text instance
‘detected’,‘ankyrin’,	split into words, trim non-alphanumeric   characters, discard stop words, and   single letters
‘ankyrin’	match as substring to name dictionary:   118 hits for ‘ankyrin’, no hit for ‘detected’
…ankyrin-3lankyrin G…	use rules to generate spelling alternatives
…ankyrin 3lankyrin-   3lankyrin Glankyrin-G…	match each alternative in full length   to the raw text
‘ …detected ankyrin G	‘ankyrin G’ matches
‘ankyrin G’, Q12955,   12507143	recognized protein name, primary   accession number, and PubMed   identifier
(Text sample from Kretschmer et al. 2002)