CoreNLP Server for Chinese coreference resolution returns empty

[Lynten]

I started the server by the command:
java -mx8g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 9000 -timeout 15000

WhenI submitted the following text: 小明吃了个冰棒，它很甜。

It returns like this:

I have checked the Chrome Developer Tools:

data sent:

response:

The corefs object is empty.

And the following command returns the same result.
java -Xmx8g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -serverProperties StanfordCoreNLP-chinese.properties -port 9000 -timeout 15000

Does CoreNLP Server support Chinese coreference resolution? If not now, will it be added in the future version?

[ryan-gilbert]

It may just be that the coref annotator couldn't identify any coreferences. It seems like the server does support Chinese coref because you received the corefs object and the Chinese properties file confirms that CoreNLP supports coref for Chinese. What are your server logs saying? Setting coref.print.md.log and coref.debug to true in that file should give you more info.

[J38]

Here is some example usage. I ran on this sentence:

习近平出生在北京。他是中国的总统。

I made a file called chinese-coref.properties:

# Pipeline options - lemma is no-op for Chinese but currently needed because coref demands it (bad old requirements system)
annotators = tokenize, ssplit, pos, lemma, ner, parse, coref

# segment
tokenize.language = zh
segment.model = edu/stanford/nlp/models/segmenter/chinese/ctb.gz
segment.sighanCorporaDict = edu/stanford/nlp/models/segmenter/chinese
segment.serDictionary = edu/stanford/nlp/models/segmenter/chinese/dict-chris6.ser.gz
segment.sighanPostProcessing = true

# sentence split
ssplit.boundaryTokenRegex = [.。]|[!?！？]+

# pos
pos.model = edu/stanford/nlp/models/pos-tagger/chinese-distsim/chinese-distsim.tagger

# ner
ner.language = chinese
ner.model = edu/stanford/nlp/models/ner/chinese.misc.distsim.crf.ser.gz
ner.applyNumericClassifiers = true
ner.useSUTime = false

# regexner
ner.fine.regexner.mapping = edu/stanford/nlp/models/kbp/chinese/cn_regexner_mapping.tab
ner.fine.regexner.noDefaultOverwriteLabels = CITY,COUNTRY,STATE_OR_PROVINCE

# parse
parse.model = edu/stanford/nlp/models/srparser/chineseSR.ser.gz

# depparse
depparse.model    = edu/stanford/nlp/models/parser/nndep/UD_Chinese.gz
depparse.language = chinese

# coref
coref.algorithm = neural
coref.language = chinese
coref.md.liberalMD = true

# kbp
kbp.semgrex = edu/stanford/nlp/models/kbp/chinese/semgrex
kbp.tokensregex = edu/stanford/nlp/models/kbp/chinese/tokensregex
kbp.language = zh
kbp.model = none

# entitylink
entitylink.wikidict = edu/stanford/nlp/models/kbp/chinese/wikidict_chinese.tsv.gz

If you start up a server, you should use the -serverProperties chinese-coref.properties option to use the properties I just specified. Obviously some of those properties are unnecessary.

When run on the example text, I get the expected coreference:

Coreference set:
        (2,1,[1,2]) -> (1,1,[1,2]), that is: "他" -> "习近平"

Please let me know if you have any other questions.

[leovan]

I still got the problem of @Lynten (run as a server).

Here is some detail of this problem:

1. CoreNLP Version: 3.9.1 and Chinese model (stanford-chinese-corenlp-2018-02-27-models.jar)
2. properties file extracted from stanford-chinese-corenlp-2018-02-27-models.jar: StanfordCoreNLP-chinese.properties

Different situations:

1. Run as cli:

CMD: java -Xmx8g -Xms4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLP -prop "StanfordCoreNLP-chinese.properties" -port 10000 -timeout 5000
INPUT: 习近平出生在北京。他是中国的总统。
OUTPUT:

......
Coreference set:
        (2,1,[1,2]) -> (1,1,[1,2]), that is: "他" -> "习近平"
......

2. Run as server:

CMD: java -Xmx8g -Xms4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -serverProperties "StanfordCoreNLP-chinese.properties" -port 10000 -timeout 5000
INPUT: 习近平出生在北京。他是中国的总统。

OUTPUT:

I also tried @J38 's properties file, it got the same results.

Since in my project, I need call the apis from python, and I use corenlp package. Some python codes are showed below:

ann = client.annotate(text, annotators='parse coref'.split())
coref_chain = ann.corefChain

with the same input of 习近平出生在北京。他是中国的总统。, it got nothing too (coref_chain == None).

Thus, I think it's not the error of the web demo project, it maybe something wrong with the server part. And I did not test with the older version to check whether they get the same problem or not.

Wish the authors can help with this problem.

[J38]

What total output are you seeing as the Java server processes the request? How does this differ from the output of the command line ? Are you building the client in Python with start_server = False. Have you tried supplying a properties dict in the annotate call? Review the annotate() method here: https://github.com/stanfordnlp/python-stanford-corenlp/blob/master/corenlp/client.py

[J38]

You should also try sending a request with a properties dict that does not specify annotators, since the annotators should be set initially by the serverProperties

[J38]

For instance I started a server with the Java command, and then set up the client in Python:

client = corenlp.CoreNLPClient(start_server=False)
ann = client.annotate(text, properties={"inputFormat":"text", "outputFormat":"serialized", 'serializer': 'edu.stanford.nlp.pipeline.ProtobufAnnotationSerializer'})

And it ran the proper Chinese pipeline and had coreference data.

[leovan]

Thanks a lot @J38 for your replying so soon. I will give more details of this problems.

1. Cli version:

Console outputs:

java -Xmx8g -Xms4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLP -prop "StanfordCoreNLP-chinese.properties" -port 10000 -timeout 5000
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator tokenize
[main] INFO edu.stanford.nlp.ie.AbstractSequenceClassifier - Loading classifier from edu/stanford/nlp/models/segmenter/chinese/ctb.gz ... done [8.5 sec].
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator ssplit
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator pos
[main] INFO edu.stanford.nlp.tagger.maxent.MaxentTagger - Loading POS tagger from edu/stanford/nlp/models/pos-tagger/chinese-distsim/chinese-distsim.tagger ... done [0.8 sec].
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator lemma
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator ner
[main] INFO edu.stanford.nlp.ie.AbstractSequenceClassifier - Loading classifier from edu/stanford/nlp/models/ner/chinese.misc.distsim.crf.ser.gz ... done [3.5 sec].
[main] WARN edu.stanford.nlp.pipeline.TokensRegexNERAnnotator - TokensRegexNERAnnotator ner.fine.regexner: Entry has multiple types for ner: 巴伐利亚 STATE_OR_PROVINCE  MISC,GPE,LOCATION       1.  Taking type to be MISC
[main] WARN edu.stanford.nlp.pipeline.TokensRegexNERAnnotator - TokensRegexNERAnnotator ner.fine.regexner: Entry has multiple types for ner: 巴伐利亚 州 STATE_OR_PROVINCE       MISC,GPE,LOCATION       1.  Taking type to be MISC
[main] INFO edu.stanford.nlp.pipeline.TokensRegexNERAnnotator - TokensRegexNERAnnotator ner.fine.regexner: Read 21238 unique entries outof 21249 from edu/stanford/nlp/models/kbp/chinese/cn_regexner_mapping.tab, 0 TokensRegex patterns.
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator parse
[main] INFO edu.stanford.nlp.parser.common.ParserGrammar - Loading parser from serialized file edu/stanford/nlp/models/srparser/chineseSR.ser.gz ... done [17.2 sec].
[main] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator coref
[main] INFO edu.stanford.nlp.pipeline.CorefMentionAnnotator - Using mention detector type: rule

Entering interactive shell. Type q RETURN or EOF to quit.
NLP> 习近平出生在北京。他是中国的总统。
Loading Chinese dictionaries from 1 file:
  edu/stanford/nlp/models/segmenter/chinese/dict-chris6.ser.gz
Done. Unique words in ChineseDictionary is: 423200.
Loading character dictionary file from edu/stanford/nlp/models/segmenter/chinese/dict/character_list [done].
Loading affix dictionary from edu/stanford/nlp/models/segmenter/chinese/dict/in.ctb [done].

Sentence #1 (5 tokens):
习近平出生在北京。

Tokens:
[Text=习近平 CharacterOffsetBegin=0 CharacterOffsetEnd=3 PartOfSpeech=NR Lemma=习近平 NamedEntityTag=PERSON]
[Text=出生 CharacterOffsetBegin=3 CharacterOffsetEnd=5 PartOfSpeech=VV Lemma=出生 NamedEntityTag=O]
[Text=在 CharacterOffsetBegin=5 CharacterOffsetEnd=6 PartOfSpeech=P Lemma=在 NamedEntityTag=O]
[Text=北京 CharacterOffsetBegin=6 CharacterOffsetEnd=8 PartOfSpeech=NR Lemma=北京 NamedEntityTag=STATE_OR_PROVINCE]
[Text=。 CharacterOffsetBegin=8 CharacterOffsetEnd=9 PartOfSpeech=PU Lemma=。 NamedEntityTag=O]

Constituency parse:
(ROOT
  (IP
    (NP (NR 习近平))
    (VP (VV 出生)
      (PP (P 在)
        (NP (NR 北京))))
    (PU 。)))


Dependency Parse (enhanced plus plus dependencies):
root(ROOT-0, 出生-2)
nsubj(出生-2, 习近平-1)
case(北京-4, 在-3)
nmod:prep(出生-2, 北京-4)
punct(出生-2, 。-5)

Extracted the following NER entity mentions:
习近平  PERSON
北京    STATE_OR_PROVINCE

Sentence #2 (6 tokens):
他是中国的总统。

Tokens:
[Text=他 CharacterOffsetBegin=9 CharacterOffsetEnd=10 PartOfSpeech=PN Lemma=他 NamedEntityTag=O]
[Text=是 CharacterOffsetBegin=10 CharacterOffsetEnd=11 PartOfSpeech=VC Lemma=是 NamedEntityTag=O]
[Text=中国 CharacterOffsetBegin=11 CharacterOffsetEnd=13 PartOfSpeech=NR Lemma=中国 NamedEntityTag=COUNTRY]
[Text=的 CharacterOffsetBegin=13 CharacterOffsetEnd=14 PartOfSpeech=DEG Lemma=的 NamedEntityTag=O]
[Text=总统 CharacterOffsetBegin=14 CharacterOffsetEnd=16 PartOfSpeech=NN Lemma=总统 NamedEntityTag=TITLE]
[Text=。 CharacterOffsetBegin=16 CharacterOffsetEnd=17 PartOfSpeech=PU Lemma=。 NamedEntityTag=O]

Constituency parse:
(ROOT
  (IP
    (NP (PN 他))
    (VP (VC 是)
      (NP
        (DNP
          (NP (NR 中国))
          (DEG 的))
        (NP (NN 总统))))
    (PU 。)))


Dependency Parse (enhanced plus plus dependencies):
root(ROOT-0, 总统-5)
nsubj(总统-5, 他-1)
cop(总统-5, 是-2)
nmod:assmod(总统-5, 中国-3)
case(中国-3, 的-4)
punct(总统-5, 。-6)

Extracted the following NER entity mentions:
中国    COUNTRY
总统    TITLE

Coreference set:
        (2,1,[1,2]) -> (1,1,[1,2]), that is: "他" -> "习近平"

2. Server version:

Request url:

http://127.0.0.1:10000/?properties=%7B%22annotators%22%3A%20%22tokenize%2Cssplit%2Cparse%2Ccoref%22%2C%20%22date%22%3A%20%222018-04-09T20%3A07%3A28%22%7D&pipelineLanguage=zh

Response:

{"docDate":"2018-04-09T20:07:28","sentences":[{"index":0,"parse":"(ROOT\n  (IP\n    (IP\n      (NP (NR 习近平))\n      (VP (VV 出生)\n        (PP (P 在)\n          (NP (NR 北京)))))\n    (PU 。)\n    (IP\n      (NP (PN 他))\n      (VP (VC 是)\n        (NP\n          (DNP\n            (NP (NR 中国))\n            (DEG 的))\n          (NP (NN 总统)))))\n    (PU 。)))","basicDependencies":[{"dep":"ROOT","governor":0,"governorGloss":"ROOT","dependent":2,"dependentGloss":"出生"},{"dep":"nsubj","governor":2,"governorGloss":"出生","dependent":1,"dependentGloss":"习近平"},{"dep":"case","governor":4,"governorGloss":"北京","dependent":3,"dependentGloss":"在"},{"dep":"nmod:prep","governor":2,"governorGloss":"出生","dependent":4,"dependentGloss":"北京"},{"dep":"punct","governor":2,"governorGloss":"出生","dependent":5,"dependentGloss":"。"},{"dep":"nsubj","governor":10,"governorGloss":"总统","dependent":6,"dependentGloss":"他"},{"dep":"cop","governor":10,"governorGloss":"总统","dependent":7,"dependentGloss":"是"},{"dep":"nmod:assmod","governor":10,"governorGloss":"总统","dependent":8,"dependentGloss":"中国"},{"dep":"case","governor":8,"governorGloss":"中国","dependent":9,"dependentGloss":"的"},{"dep":"conj","governor":2,"governorGloss":"出生","dependent":10,"dependentGloss":"总统"},{"dep":"punct","governor":2,"governorGloss":"出生","dependent":11,"dependentGloss":"。"}],"enhancedDependencies":[{"dep":"ROOT","governor":0,"governorGloss":"ROOT","dependent":2,"dependentGloss":"出生"},{"dep":"nsubj","governor":2,"governorGloss":"出生","dependent":1,"dependentGloss":"习近平"},{"dep":"case","governor":4,"governorGloss":"北京","dependent":3,"dependentGloss":"在"},{"dep":"nmod:prep","governor":2,"governorGloss":"出生","dependent":4,"dependentGloss":"北京"},{"dep":"punct","governor":2,"governorGloss":"出生","dependent":5,"dependentGloss":"。"},{"dep":"nsubj","governor":10,"governorGloss":"总统","dependent":6,"dependentGloss":"他"},{"dep":"cop","governor":10,"governorGloss":"总统","dependent":7,"dependentGloss":"是"},{"dep":"nmod:assmod","governor":10,"governorGloss":"总统","dependent":8,"dependentGloss":"中国"},{"dep":"case","governor":8,"governorGloss":"中国","dependent":9,"dependentGloss":"的"},{"dep":"conj","governor":2,"governorGloss":"出生","dependent":10,"dependentGloss":"总统"},{"dep":"punct","governor":2,"governorGloss":"出生","dependent":11,"dependentGloss":"。"}],"enhancedPlusPlusDependencies":[{"dep":"ROOT","governor":0,"governorGloss":"ROOT","dependent":2,"dependentGloss":"出生"},{"dep":"nsubj","governor":2,"governorGloss":"出生","dependent":1,"dependentGloss":"习近平"},{"dep":"case","governor":4,"governorGloss":"北京","dependent":3,"dependentGloss":"在"},{"dep":"nmod:prep","governor":2,"governorGloss":"出生","dependent":4,"dependentGloss":"北京"},{"dep":"punct","governor":2,"governorGloss":"出生","dependent":5,"dependentGloss":"。"},{"dep":"nsubj","governor":10,"governorGloss":"总统","dependent":6,"dependentGloss":"他"},{"dep":"cop","governor":10,"governorGloss":"总统","dependent":7,"dependentGloss":"是"},{"dep":"nmod:assmod","governor":10,"governorGloss":"总统","dependent":8,"dependentGloss":"中国"},{"dep":"case","governor":8,"governorGloss":"中国","dependent":9,"dependentGloss":"的"},{"dep":"conj","governor":2,"governorGloss":"出生","dependent":10,"dependentGloss":"总统"},{"dep":"punct","governor":2,"governorGloss":"出生","dependent":11,"dependentGloss":"。"}],"entitymentions":[{"docTokenBegin":0,"docTokenEnd":1,"tokenBegin":0,"tokenEnd":1,"text":"习近平","characterOffsetBegin":0,"characterOffsetEnd":3,"ner":"PERSON"},{"docTokenBegin":3,"docTokenEnd":4,"tokenBegin":3,"tokenEnd":4,"text":"北京","characterOffsetBegin":6,"characterOffsetEnd":8,"ner":"STATE_OR_PROVINCE"},{"docTokenBegin":7,"docTokenEnd":8,"tokenBegin":7,"tokenEnd":8,"text":"中国","characterOffsetBegin":11,"characterOffsetEnd":13,"ner":"COUNTRY"},{"docTokenBegin":9,"docTokenEnd":10,"tokenBegin":9,"tokenEnd":10,"text":"总统","characterOffsetBegin":14,"characterOffsetEnd":16,"ner":"TITLE"}],"tokens":[{"index":1,"word":"习近平","originalText":"习近平","lemma":"习近平","characterOffsetBegin":0,"characterOffsetEnd":3,"pos":"NR","ner":"PERSON","speaker":"PER0"},{"index":2,"word":"出生","originalText":"出生","lemma":"出生","characterOffsetBegin":3,"characterOffsetEnd":5,"pos":"VV","ner":"O","speaker":"PER0"},{"index":3,"word":"在","originalText":"在","lemma":"在","characterOffsetBegin":5,"characterOffsetEnd":6,"pos":"P","ner":"O","speaker":"PER0"},{"index":4,"word":"北京","originalText":"北京","lemma":"北京","characterOffsetBegin":6,"characterOffsetEnd":8,"pos":"NR","ner":"STATE_OR_PROVINCE","speaker":"PER0"},{"index":5,"word":"。","originalText":"。","lemma":"。","characterOffsetBegin":8,"characterOffsetEnd":9,"pos":"PU","ner":"O","speaker":"PER0"},{"index":6,"word":"他","originalText":"他","lemma":"他","characterOffsetBegin":9,"characterOffsetEnd":10,"pos":"PN","ner":"O","speaker":"PER0"},{"index":7,"word":"是","originalText":"是","lemma":"是","characterOffsetBegin":10,"characterOffsetEnd":11,"pos":"VC","ner":"O","speaker":"PER0"},{"index":8,"word":"中国","originalText":"中国","lemma":"中国","characterOffsetBegin":11,"characterOffsetEnd":13,"pos":"NR","ner":"COUNTRY","speaker":"PER0"},{"index":9,"word":"的","originalText":"的","lemma":"的","characterOffsetBegin":13,"characterOffsetEnd":14,"pos":"DEG","ner":"O","speaker":"PER0"},{"index":10,"word":"总统","originalText":"总统","lemma":"总统","characterOffsetBegin":14,"characterOffsetEnd":16,"pos":"NN","ner":"TITLE","speaker":"PER0"},{"index":11,"word":"。","originalText":"。","lemma":"。","characterOffsetBegin":16,"characterOffsetEnd":17,"pos":"PU","ner":"O","speaker":"PER0"}]}],"corefs":{}}

Console output:

java -Xmx8g -Xms4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -serverProperties "StanfordCoreNLP-chinese.properties" -port 10000 -timeout 5000
[main] INFO CoreNLP - --- StanfordCoreNLPServer#main() called ---
[main] INFO CoreNLP - setting default constituency parser
[main] INFO CoreNLP - warning: cannot find edu/stanford/nlp/models/srparser/englishSR.ser.gz
[main] INFO CoreNLP - using: edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz instead
[main] INFO CoreNLP - to use shift reduce parser download English models jar from:
[main] INFO CoreNLP - http://stanfordnlp.github.io/CoreNLP/download.html
[main] INFO CoreNLP -     Threads: 4
[main] INFO CoreNLP - Starting server...
[main] INFO CoreNLP - StanfordCoreNLPServer listening at /0:0:0:0:0:0:0:0:10000
[pool-1-thread-1] INFO CoreNLP - [/127.0.0.1:53541] API call w/annotators tokenize,ssplit,pos,lemma,ner,parse,coref
习近平出生在北京。他是中国的总统。
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator tokenize
[pool-1-thread-1] INFO edu.stanford.nlp.ie.AbstractSequenceClassifier - Loading classifier from edu/stanford/nlp/models/segmenter/chinese/ctb.gz ... done [7.0 sec].
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator ssplit
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator pos
[pool-1-thread-1] INFO edu.stanford.nlp.tagger.maxent.MaxentTagger - Loading POS tagger from edu/stanford/nlp/models/pos-tagger/chinese-distsim/chinese-distsim.tagger ... done [0.9 sec].
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator lemma
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator ner
[pool-1-thread-1] INFO edu.stanford.nlp.ie.AbstractSequenceClassifier - Loading classifier from edu/stanford/nlp/models/ner/chinese.misc.distsim.crf.ser.gz ... done [3.4 sec].
[pool-1-thread-1] WARN edu.stanford.nlp.pipeline.TokensRegexNERAnnotator - TokensRegexNERAnnotator ner.fine.regexner: Entry has multipletypes for ner: 巴伐利亚 STATE_OR_PROVINCE       MISC,GPE,LOCATION       1.  Taking type to be MISC
[pool-1-thread-1] WARN edu.stanford.nlp.pipeline.TokensRegexNERAnnotator - TokensRegexNERAnnotator ner.fine.regexner: Entry has multipletypes for ner: 巴伐利亚 州 STATE_OR_PROVINCE    MISC,GPE,LOCATION       1.  Taking type to be MISC
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.TokensRegexNERAnnotator - TokensRegexNERAnnotator ner.fine.regexner: Read 21238 unique entries out of 21249 from edu/stanford/nlp/models/kbp/chinese/cn_regexner_mapping.tab, 0 TokensRegex patterns.
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator parse
[pool-1-thread-1] INFO edu.stanford.nlp.parser.common.ParserGrammar - Loading parser from serialized file edu/stanford/nlp/models/srparser/chineseSR.ser.gz ... done [16.6 sec].
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator coref
[pool-1-thread-1] INFO edu.stanford.nlp.pipeline.CorefMentionAnnotator - Using mention detector type: rule
[pool-2-thread-1] INFO edu.stanford.nlp.wordseg.ChineseDictionary - Loading Chinese dictionaries from 1 file:
[pool-2-thread-1] INFO edu.stanford.nlp.wordseg.ChineseDictionary -   edu/stanford/nlp/models/segmenter/chinese/dict-chris6.ser.gz
[pool-2-thread-1] INFO edu.stanford.nlp.wordseg.ChineseDictionary - Done. Unique words in ChineseDictionary is: 423200.
[pool-2-thread-1] INFO edu.stanford.nlp.wordseg.CorpusChar - Loading character dictionary file from edu/stanford/nlp/models/segmenter/chinese/dict/character_list [done].
[pool-2-thread-1] INFO edu.stanford.nlp.wordseg.AffixDictionary - Loading affix dictionary from edu/stanford/nlp/models/segmenter/chinese/dict/in.ctb [done].

3. Python client:

1). I do build the client in Python with start_server = False (use the server started in console above).
2). Python codes:

ann = client.annotate(text, annotators='parse coref'.split(), properties={"inputFormat":"text", "outputFormat":"serialized", 'serializer': 'edu.stanford.nlp.pipeline.ProtobufAnnotationSerializer'})
print(ann)
coref_chain = ann.corefChain
print(coref_chain)

3). Console outputs:

[pool-1-thread-3] INFO CoreNLP - [/127.0.0.1:53274] API call w/annotators tokenize,ssplit,pos,lemma,ner,parse,coref
习近平出生在北京，他是中国的总统。
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator tokenize
[pool-1-thread-3] INFO edu.stanford.nlp.ie.AbstractSequenceClassifier - Loading classifier from edu/stanford/nlp/models/segmenter/chinese/ctb.gz ... done [7.3 sec].
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator ssplit
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator pos
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator lemma
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator ner
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator parse
[pool-1-thread-3] INFO edu.stanford.nlp.pipeline.StanfordCoreNLP - Adding annotator coref
[pool-2-thread-2] INFO edu.stanford.nlp.wordseg.ChineseDictionary - Loading Chinese dictionaries from 1 file:
[pool-2-thread-2] INFO edu.stanford.nlp.wordseg.ChineseDictionary -   edu/stanford/nlp/models/segmenter/chinese/dict-chris6.ser.gz
[pool-2-thread-2] INFO edu.stanford.nlp.wordseg.ChineseDictionary - Done. Unique words in ChineseDictionary is: 423200.
[pool-2-thread-2] INFO edu.stanford.nlp.wordseg.CorpusChar - Loading character dictionary file from edu/stanford/nlp/models/segmenter/chinese/dict/character_list [done].
[pool-2-thread-2] INFO edu.stanford.nlp.wordseg.AffixDictionary - Loading affix dictionary from edu/stanford/nlp/models/segmenter/chinese/dict/in.ctb [done].

4). Since the output of print() in Python is too long, I attached the txt file instead.
test_corenlp_outputs.txt
5). I also notice that in the result of python, it has some key named mentionsForCoref, and I'm not quite sure what it is.

[J38]

Try running it in Python without the annotators option in the annotate() call as I did in my example Python. When I did that I was seeing coref results show up.

That is:

change:
ann = client.annotate(text, annotators='parse coref'.split(), properties={"inputFormat":"text", "outputFormat":"serialized", 'serializer': 'edu.stanford.nlp.pipeline.ProtobufAnnotationSerializer'})

to:
ann = client.annotate(text, properties={"inputFormat":"text", "outputFormat":"serialized", 'serializer': 'edu.stanford.nlp.pipeline.ProtobufAnnotationSerializer'})

For instance in your console output, you should see the same output for the pipeline being built as you do for the command line version. As I said above I ran that Python code and I successfully saw coref data come back for Chinese, so that makes me think the issue is you adding the annotators= parameter.

[leovan]

Hi, @J38 , I think I have figured out what's wrong with my codes.

1. In my python codes, text I posted in the issue is 习近平出生在北京。他是中国的总统。，but in my python codes it's 习近平出生在北京，他是中国的总统。. Difference is the comma between two sentences. After I corrected that, I can get the correct ann.corefChain result with or without annotators='parse coref'.split().
2. The web demo still has this problem, I tried with English sentence, it works fine.
3. I don't have a deep dive into the coreference algorithms in CoreNLP, so I am not sure that changing the period to comma gets different result is a bug or not.

Thanks for your patience with my problems.

[Lynten]

I agree with @leovan , the web demo still return an empty coref result even if I launch the server with chinese-coref.properties.

[Hacken-L]

@J38 In your file chinese-coref.properties, if use the latest Chinese model 3.9.2, there will be a mistake:

Exception in thread "main" edu.stanford.nlp.io.RuntimeIOException: Couldn't read TokensRegexNER from edu/stanford/nlp/models/kbp/chinese/cn_regexner_mapping.tab

if regexner like:

regexner

ner.fine.regexner.mapping = edu/stanford/nlp/models/kbp/chinese/cn_regexner_mapping.tab

It will be fine if change the file into

regexner

ner.fine.regexner.mapping = edu/stanford/nlp/models/kbp/chinese/gazetteers/cn_regexner_mapping.tab