I am trying to do some co-referencing using Stanford Corenlp version 4.5.4 using Python with the library Stanza latest version (1.5.0), Stanza turns on the server with the properties that I choose, these are the properties I used:
'annotators': 'tokenize,ssplit,coref'
'coref.algorithm': 'neural'
'coref.postprocessing.rules': 'true'
'dcoref.animate': 'true'
'dcoref.inanimate': 'true'
'dcoref.gender': 'unknown'
'ner.useSUTime': 'false'
after using these properties, I run the code and for some reason I get this error from the server (run in Java not python), The class
edu.stanford.nlp.hcoref.md.MentionDetectionClassifier
was not found I tried other versions of Stanza and other versions of Stanford corenlp, I always keep getting the same error, I even tried adding this property:
coref.mentionDetectionClassifier:edu.stanford.nlp.coref.neural.DocCorefNeuralMentionClassifier
but made no difference. I researched a bit on the internet, and I found out that the MentionDetectionClassifier class exists but with "coref" instead of "hcoref" in my current corenlp version (4.5.4), "hcoref" only exists in 3.6.0 stanford corenlp and below. I have no idea why this is happening. Perhaps I did something wrong with the properties?
of course if I remove the property "coref.algorithm", it uses the default value which is rule, and does not use neural networks during the co-referencing process, which results in many errors.