Stanford Return all matched expressions - stanford-nlp

Question

Is there a way to return all the matched expressions?
Consider the following sentence
John Snow killed Ramsay Bolton
where
John-NNP, Snow-NNP, killed- VBD, Ramsay-NNP, Bolton-NNP
And I am using the following tag combination as rules
NNP-NNP
NNP-VBD
VBD-NNP
and expected matched words from the above rules are:
John Snow, Snow killed, killed Ramsay, Ramsay Bolton
But using the below code, I am getting only this as matched expression:
[John Snow, killed Ramsay]
Is there a way in stanford to get all the expected matching words from the sentence? This is the code and rule file I am using right now:
import com.factweavers.multiterm.SetNLPAnnotators;
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.ling.tokensregex.CoreMapExpressionExtractor;
import edu.stanford.nlp.ling.tokensregex.Env;
import edu.stanford.nlp.ling.tokensregex.NodePattern;
import edu.stanford.nlp.ling.tokensregex.TokenSequencePattern;
import edu.stanford.nlp.pipeline.Annotation;
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
import edu.stanford.nlp.util.CoreMap;
import java.util.List;
import java.util.regex.Pattern;
public class StanfordTest {
public static void main(String[] args) {
String rulesFile="en.rules";
Env env = TokenSequencePattern.getNewEnv();
env.setDefaultStringMatchFlags(NodePattern.NORMALIZE);
env.setDefaultStringPatternFlags(Pattern.CASE_INSENSITIVE);
env.bind("collapseExtractionRules", false);
CoreMapExpressionExtractor extractor= CoreMapExpressionExtractor.createExtractorFromFiles(env, rulesFile);
String content="John Snow killed Ramsay Bolton";
Annotation document = new Annotation(content);
SetNLPAnnotators snlpa = new SetNLPAnnotators();
StanfordCoreNLP pipeline = snlpa.setAnnotators("tokenize, ssplit, pos, lemma, ner");
pipeline.annotate(document);
List<CoreMap> sentences = document.get(CoreAnnotations.SentencesAnnotation.class);
sentences.parallelStream().forEach(sentence -> {
System.out.println(extractor.extractExpressions(sentence));
});
}
}
en.rules
{
ruleType:"tokens",
pattern:([{tag:/VBD/}][ {tag:/NNP/}]),
result:"result1"
}
{
ruleType:"tokens",
pattern:([{tag:/NNP/}][ {tag:/VBD/}]),
result:"result2"
}
{
ruleType:"tokens",
pattern:([{tag:/NNP/}][ {tag:/NNP/}]),
result:"result3"
}

Answer 1

I think you need to create different extractors for different things you want.
The issue here is that when you have two part-of-speech tag rule sequences that overlap like this, the first one that gets matched absorbs the tokens preventing the second pattern from matching.
So if (NNP, NNP) is the first rule, "John Snow" gets matched. But then the Snow is not available to be matched with "Snow killed".
If you have a set of patterns that overlap like this, you should disentangle them and put them in separate extractors.
So you can have a (noun, verb) extractor, and a separate (noun, noun) extractor for instance.