GeneTaggerCRF: An Entity Tagger for Recognizing Gene Names in Text

Summary: GeneTaggerCRF is an application for tagging gene names and references in text. It is based on the MALLET natural language processing package and uses a machine learning technique called conditional random fields. It is the first installment of a set of information extraction tools called BioSFIER (Biology Software For Information Extraction and Retrieval). We tested GeneTaggerCRF's performance by 10- fold cross validation on 190 MEDLINE abstracts pertaining to Neuroblastoma cancer genetics. Performance was 0.93 precision and 0.60 recall. Availability: The software is available at