There is no official information about brat automatic annotation services. I just did not found any example about relations (so i proposed something like:"R1": {"Arg1": "T5", "Arg2": "T3", "label": "relname"})
However, +1 if opennlp includes brat support. By the way, I coded a pyhton Class that deals with brat convertionts: it includes static methods to go from .ann to python json, .ann to csv, string (formated ann) to dict etc: https://bitbucket.org/conabio_cmd/text-mining/src/80e27036ceaad31ef9e3df12fa7a37c557e23bf2/txtmining/txtmining/converters/brat.py?at=dev&fileviewer=file-view-default -- Alejandro Molina Villegas Ecoinformática, CONABIO, México On Wed, Oct 19, 2016 at 3:10 PM, Joern Kottmann <kottm...@gmail.com> wrote: > Don't know if there is an documentation, it is a JSON format, I had some > problems to get that working and then brat people fixed things a bit. > > Anyway, it is rather simple, have a look at the code, I don't have any > better reference for you. > > https://github.com/apache/opennlp-sandbox/blob/master/ > opennlp-brat-annotator/src/main/java/opennlp/bratann/ > NameFinderResource.java > > It gets called with the model name and text, and then return Map<String, > NameAnn> as JSON. > > Jörn > > On Wed, Oct 19, 2016 at 9:36 PM, Richard Eckart de Castilho < > richard.eck...@gmail.com> wrote: > > > Sorry, I meant protocol, not format... I do know the brat format. > > > > Best, > > > > -- Richard > > > > > On 19.10.2016, at 21:30, Joern Kottmann <kottm...@gmail.com> wrote: > > > > > > Have a look at this page: > > > http://brat.nlplab.org/standoff.html > > > > > > Jörn > > > > > > > > > On Wed, Oct 19, 2016 at 9:06 PM, Richard Eckart de Castilho < > > > richard.eck...@gmail.com> wrote: > > > > > >> On 19.10.2016, at 20:59, Joern Kottmann <kottm...@gmail.com> wrote: > > >>> > > >>> There is a dedicated servlet which implements exactly the protocol > brat > > >>> requires. We can extend it to make it available for other tools. > > >> > > >> Slightly off topic: is that format documented somewhere? I was looking > > at > > >> the brat documentation, but didn't find much info really. > > >> > > >> Best, > > >> > > >> -- Richard > > >> > > > > >