Example of segmentation of a phenotype description according to the two phases of our approach. In phase I, tokens are labeled using BIO labeling corresponding to each class: Q (quality), A (anatomy), C (connective). Label O (not present in the figure) is used to denote tokens outside the target classes. Similarly, in phase II, specific classifiers are used to segment quality-qualifier pairs (Q-QF) and anatomical parts (A), sub-parts (AP) and coordinates (P).
Groza et al. BMC Bioinformatics 2012 13:265 doi:10.1186/1471-2105-13-265