Currently NEAM does not do any coreference resolution among found entities. For this reason, ref attributes are based solely on the individual tokens. This does not follow the rules of the tagging instructions of the project which can be found here: https://github.com/eba-diary/EBA-xml-TEI/wiki . With coreference resolution, the system would be better able to assign the ref vales for entities (for example, providing a last name in the ref attribute even if that particular instance of the entity doesn't have a last name).
Coreference resolution is included in Stanford CoreNLP (link). It would be trivial to enable it; integrating the coreference information would be the hard part.
Currently NEAM does not do any coreference resolution among found entities. For this reason, ref attributes are based solely on the individual tokens. This does not follow the rules of the tagging instructions of the project which can be found here: https://github.com/eba-diary/EBA-xml-TEI/wiki . With coreference resolution, the system would be better able to assign the ref vales for entities (for example, providing a last name in the ref attribute even if that particular instance of the entity doesn't have a last name).