UIMA/dkpro: Get type of conjunction

Question

I am using UIMA in combination with UIMAfit and dkpro and the StanfordParser to parse english sentences.

I can build dependency trees without any problem. For "and"/"or" conjunctions, I get a Annotation with class CONJ, which is a subclass of Dependency. As of now, I did not find out, how to check if the found conjunction is an "AND" or an "OR" conjunction.

Does anybody know how to solve this? I saw examples where the dependencies "conj_and" and "conj_or" are displayed, but I dont see where they come from. http://nlp.stanford.edu/software/dependencies_manual.pdf

Thanks in advance

Some code for visualization:

// CONJ a;
// StringBuilder sb
Token dependent = a.getDependent();
Token governor = a.getGovernor();

sb.append("Dependent: ");
sb.append(dependent);
sb.append(", ");

sb.append("Governor: ");
sb.append(governor);

// How to check type conj_and/conj_or?

Answer 1

The method to get the label of the dependency relation is called

getDependencyType()

Mind that the conj_or is a "collapsed dependency". If you want to get these, you have to explicitly set the dependency mode when invoking the DKPro Core StanfordParser component, e.g. using COLLAPSED or maybe CC_PROPAGATED:

AnalysisEngineFactory.createEngineDescription(StanfordParser.class,
  StanfordParser.PARAM_MODE, StanfordParser.DependenciesMode.COLLAPSED)

The default setting for this parameter is TREE.

See also:

• DKPro Core Stanford.Parser.DEpendenciesMode Javadoc
• Stanford CoreNLP GrammaticalStructure Javadoc

Disclosure: I am a member of the DKPro Core team.