The normal ANN run sequence may be found in run.lisp.
It may be invoked either from the REPL or shell: e.g. sbcl --load run.lisp.
If you want to run web app, be sure to add :dev into *features* like rlwrap sbcl --eval "(push :dev *features*)" --load run.lisp or in REPL.
The documents for annotation should be put into the data/ dir. Currently, a small portion of the Enron corpus is
To select a particular annotation scheme the root folder of the texts to annotate should have a file .ann.yaml (similar in format to the default one).
New annotation schemes may be defined in schemas/ dir. The default one is for NER annotation.
The annotation process is intuitive: just select the spans to annotate and choose the appropriate class. Incorrect annotations may be removed. All changes are persisted to disk in an .ann file alongside the text file at once.
There's ability to setup basic auth for users, for that place users.txt with each line being %username% %password%
The license of ANN is Apache 2.0.
(c) 2016-2017, Vsevolod Dyomkin [email protected]