For example, the Phenex tool was designed to support the phenotype annotation of character matrices in the Phenoscape project [ 48].
Annotations, white space characters, newline characters and block delimiters are some of the most common culprits of wasted code space.
EQ annotation of characters (both datasets) and taxa (FC dataset only) with anatomy, quality and taxonomy ontologies was done using Phenex (4, 6).
The completion of the FC dataset, taking data preparation, annotation of characters and taxa, ontology development, and team discussions into account, required a total of two person-years (4136 h), with a resulting overall annotation rate of 0.65 characters per hour.
The almost daily updated documentation of annotation problems, examples, and standards in the Guide to Character Annotation [69] was important in promoting these annotation standards.
For the second dataset, termed the character annotation only (CA) dataset, only the character annotation task (ii above) was done in the context of an inter-curator annotation experiment (Manda et al., in preparation).
Considering only the character annotation component (1467 h) of the FC dataset yields an annotation rate of 1.84 characters per hour.
The large overhead required for data preparation and ontology development contributed to a low annotation rate of approximately two characters per hour, compared with 14 characters per hour when activity was restricted to character annotation.
If a more stringent filter is applied, by removing annotations with three characters or less, recall drops by 12.6 percentage points while precision only improves slightly, by 2.6%.
Our proposed method is based on character annotation and is not the only strategy that can be used to discover character relationships and the appearances of characters "on-screen" visually.
