dkpro
Add support for DKPro type system and annotations
it would be nice to be able to initialize a CAS with a certain type system, e.g.
from somewhere import DKProCoreTypeSystem
from cassis import Cas
cas = Cas(DKProCoreTypeSystem())
In order to verify that the DKPro typeystem is supported by cassis, the typesystem xml and an example CAS should be added to the tests.
Here is a single type system file for a start which includes most DKPro Core types (not all, but all important ones - i.e. those from the API modules that are part of DKPro Core 1.10.0):
Right now, it fails because uima.tcas.DocumentAnnotation is missing. Is it also a predefined annotation like uima.tcas.DocumentAnnotation and uima.tcas.DocumentAnnotation?
@Rentier I have tried extracting most of the "internal" type system of UIMA. Mind this is only an indicative type system XML file to give you an easier overview of how UIMA initializes the type system of an empty CAS. The file itself can actually not be loaded with Java UIMA - UIMA would complain during loading.
Note that the TS does not include any built-in types which do not have a parent type. That's specifically the uima.cas.TOP type.
I found that when I serialize an UIMA Cas then it always creates a uima.tcas.DocumentAnnotation in both cas and type system. Therefore I do not add it to the predefined types.
I added the type system support, I wait for @reckart to give me some CAS XMI that contains one of each DKPro annotation (or as many different types as possible).
Just played around with this module (v0.2.7) and tried to use load_dkpro_core_typesystem() but it seems that "dkpro-core-types.xml" is absent from the PyPI package (see https://pypi.org/project/dkpro-cassis/#files), causing the function to tail. Adding in the XML file from the repo fixed it but you might want to look at this to make the released version work.
Thanks for this neat module!
