MAT 2.0 introduces a new, extremely detailed mechanism for
describing the annotation types available to a given task. The
documented description language is XML.
However, the compiled information is made available to the MAT UI
as JSON, and this JSON format
is also used in the MAT standalone viewer/annotator and the Java
library. This tool converts the information in a task into the
appropriate JSON format, as long as the task has been installed
using MATManagePluginDirs.
Unix:
% $MAT_PKG_HOME/bin/MATAnnotationInfoToJSON
Windows native:
> %MAT_PKG_HOME%\bin\MATAnnotationInfoToJSON
Usage: MATAnnotationInfoToJSON [options] task outfile
task: the name of a MAT task.
outfile: the output file to write the JSON version of the annotation info to. If the file
is '-', the JSON will be written to standard output.
The tool also allows you to control whether redundant info is
displayed; whether the output is pretty-printed or not; and
whether noncontent annotations are produced.
| --task_directory <dir> |
An additional task directory to load to find
the task in. |
| --dont_remove_redundant_info |
By default, the tool removes
redundant information to improve readability. If this flag
is present, the redundant information will be retained. Both
the Java library and the Javascript standalone viewer are
configured to repopulate redundant information. |
| --compact |
By default, the tool
pretty-prints its JSON output. If this flag is present, the
JSON will not be pretty-printed |
| --keep_noncontent_annotations |
By default, the tool does not
preserve token, zone or admin annotation type descriptors.
If this flag is present, all annotation types will be
presented. |
| --simplified |
By default, the tool dumps expanded
JSON, in which all the annotation-related task
information, including tag order, hierarchy, etc., even if
this information is not present. If this flag is present,
the tool will dump simplified
JSON, which is just a list of the annotation types. |
Let's say you want to convert the included "Named Entity" task
and save the content annotations to the file ne.json,
pretty-printed, without redundant information:
Unix:
% $MAT_PKG_HOME/bin/MATAnnotationInfoToJSON "Named Entity" ne.json
Windows native:
> %MAT_PKG_HOME%\bin\MATAnnotationInfoToJSON.cmd "Named Entity" ne.json
Like example 1, but you want the noncontent annotations as well,
and you want the output printed to standard output:
Unix:
% $MAT_PKG_HOME/bin/MATAnnotationInfoToJSON --keep_noncontent_annotations "Named Entity" -
Windows native:
> %MAT_PKG_HOME%\bin\MATAnnotationInfoToJSON.cmd --keep_noncontent_annotations "Named Entity" -