Hi all, I spent a bit of time on the Tagging Server. For now the primary reason to make it is to have the ability to access OpenNLP from browser based clients.
Such clients could be an online demo, annotation tools, browser plugins, etc. The architecture is very simple, its based on OSGi and just exposes the OpenNLP API via REST. Models are assumed to be available as a service in OSGi. They can easily be registered via blueprint (dependency injection) and loaded from a URL (file, http, ...). More sophisticated providers are possible, e.g a bundle which trains a model. The models are cached and the components are created per request, I am not sure if that is too expensive but its a super easy way to make everything thread safe. I will add some instructions on how to get it running to the project. The code for the name finder and pos tagger is checked in now, please have look, any comments are welcome. Jörn
