View article

For the recognition of gene and protein names and their normalization to gene and protein centered databases (Entrez Gene and UniProt) regularly updated dictionaries generated from these sources are used by the ProMiner system to search gene and protein names in scientific publications. A multistep curation process and inclusion of different biomedical dictionaries in the curation process leads to an increase of precision and recall. The recognition of names containing special parenthesis expressions augments the recall further. Human gene and protein names in the test corpus provided in BioCreAtIvE II could be recognized with the adapted ProMiner system and a regularly updated dictionary with a final F-measure of 80%.