Exploiting the contextual cues for bio-entity name recognition in biomedical literature

To extract biomedical information about bio-entities from the huge amount of biomedical literature, the first key step is recognizing their names in these literatures, which remains a challenging task due to the irregularities and ambiguities in bio-entities nomenclature. The recognition performances of the current popular methods, machine learning techniques, still have much space to be improved. This paper presents a Conditional Random Field-based approach used to recognize the names of bio-entities including gene, protein, cell type, cell line and studies the methods of improving the performance by the exploitation of the contextual cues including bracket pair, heuristic syntax structure and interaction words cue. Experiment results on both JNLPBA2004 and BioCreative2004 task 1A datasets show that these methods can improve Conditional Random Field-based recognition performance by more than 2 points in F-score.