Closed aveekchoudhury01 closed 5 years ago
The coreference function only works for coreNLP, unfortunately. I think the spacy people
have a coreference resolution system in the works and I'll port it over as soon as it seems
usable. I'm not sure why coreNLP is not working, however, could you provide a working
example? Note that you need to manually set the anno_level
parameter in cnlp_init_corenlp
to 3 or greater for the co-reference resolution annotator to get turned on.
And I too have found that the coreNLP coreference resolution is incredibly slow. You may need to increase the amount of memory assigned to rjava if it constantly hangs.
There is now a working coreference resolution system in spacy that is quite effective. Are there plans to port it over to cleannlp? Unfortunately, the CoreNLP coref is too slow unless you're working with a large computer cluster of some sort.
Hi Taylor, I am using the cleanNLP package to address co-reference resolution. Unfortunately for me, the cnlp_get_coreference() returns 0 rows of data. Following are the examples I ran -
Machine config - "Ubuntu 16.04", 32GB RAM
Steps followed -
Can you help out as to why these results?
Thanks, Aveek