Abstract
BACKGROUND: The identification of mentions of gene or gene products in biomedical texts is a critical step in the development of text mining applications in biosciences. The complexity and ambiguity of gene nomenclature makes this a very difficult task. METHODS: Here we present a novel approach based on a combination of carefully designed rules and several lexicons of biological concepts, implemented in the Text Detective system. Text Detective is able to normalize the results of gene mentions found by offering the appropriate database reference. RESULTS: In BioCreAtIvE evaluation, Text Detective achieved results of 84% precision, 71% recall for task 1A, and 79% precision, 71% recall for mouse genes in task 1B.