Provides support for loading/saving annotations stored in the brat standoff format
To load documents in the brat format make sure you have the .ann file next to
the .txt file and that, extension excepting, the files share the same name.
Now load the .txt file into GATE in any of the normal ways specifying
text/x-brat as the MIME type.
You can also load extra sets of brat format annotations onto an existing document through the GUI by right clicking on a document and choosing Import brat Annotations.... This allows you to select a file of brat standoff annotations and the annotation set into which they should be imported.
To save documets in the brat format, right click on the document or corpus, and
select the brat Standoff Annotations option. This will save just the
annotations (i.e. it will produce the .ann file) with offsets relative to the
document as visible in GATE. This means that if the document wasn't created
from a plain text file the offsets will be dependent on how the relevant GATE
document format unpacked the original document.
brat annotations can include normalization information (see the docs for more details) which link annotations to entities in a remote knowledge base; the examples link annotations to Wikipedia entries for example.
The normalization information present in the .ann files is preserved upon
document loading, but is left in it's raw encoded form. The brat Noramlizer
PR can be used to expand this information into full URLs using the details
found in the relevant tools.conf file.
The PR can also be configured to collapse the full URLs back to their encoded
form ready for saving as brat annotations again should this be necessary using
the type parameter.