/// <summary>Annotate a document (which is usually just a sentence).</summary>
public virtual void Annotate(StanfordCoreNLP pipeline, Annotation ann)
{
if (ann.Get(typeof(CoreAnnotations.SentencesAnnotation)) == null)
{
pipeline.Annotate(ann);
}
else
{
if (ann.Get(typeof(CoreAnnotations.SentencesAnnotation)).Count == 1)
{
ICoreMap sentence = ann.Get(typeof(CoreAnnotations.SentencesAnnotation))[0];
foreach (CoreLabel token in sentence.Get(typeof(CoreAnnotations.TokensAnnotation)))
{
token.Remove(typeof(NaturalLogicAnnotations.OperatorAnnotation));
token.Remove(typeof(NaturalLogicAnnotations.PolarityAnnotation));
}
sentence.Remove(typeof(NaturalLogicAnnotations.RelationTriplesAnnotation));
sentence.Remove(typeof(NaturalLogicAnnotations.EntailedSentencesAnnotation));
sentence.Remove(typeof(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation));
sentence.Remove(typeof(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation));
sentence.Remove(typeof(SemanticGraphCoreAnnotations.EnhancedPlusPlusDependenciesAnnotation));
pipeline.Annotate(ann);
}
}
}
/// <summary>Annotate a single sentence.</summary>
/// <remarks>
/// Annotate a single sentence.
/// This annotator will, in particular, set the
/// <see cref="EntailedSentencesAnnotation"/>
/// and
/// <see cref="RelationTriplesAnnotation"/>
/// annotations.
/// </remarks>
public virtual void AnnotateSentence(ICoreMap sentence, IDictionary <CoreLabel, IList <CoreLabel> > canonicalMentionMap)
{
IList <CoreLabel> tokens = sentence.Get(typeof(CoreAnnotations.TokensAnnotation));
if (tokens.Count < 2)
{
// Short sentence. Skip annotating it.
sentence.Set(typeof(NaturalLogicAnnotations.RelationTriplesAnnotation), Java.Util.Collections.EmptyList());
if (!stripEntailments)
{
sentence.Set(typeof(NaturalLogicAnnotations.EntailedSentencesAnnotation), Java.Util.Collections.EmptySet());
}
}
else
{
// Get the dependency tree
SemanticGraph parse = sentence.Get(typeof(SemanticGraphCoreAnnotations.EnhancedPlusPlusDependenciesAnnotation));
if (parse == null)
{
parse = sentence.Get(typeof(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation));
}
if (parse == null)
{
throw new InvalidOperationException("Cannot run OpenIE without a parse tree!");
}
// Clean the tree
parse = new SemanticGraph(parse);
Edu.Stanford.Nlp.Naturalli.Util.CleanTree(parse);
// Resolve Coreference
SemanticGraph canonicalizedParse = parse;
if (resolveCoref && !canonicalMentionMap.IsEmpty())
{
canonicalizedParse = CanonicalizeCoref(parse, canonicalMentionMap);
}
// Run OpenIE
// (clauses)
IList <SentenceFragment> clauses = ClausesInSentence(canonicalizedParse, true);
// note: uses coref-canonicalized parse
// (entailment)
ICollection <SentenceFragment> fragments = EntailmentsFromClauses(clauses);
// (segment)
IList <RelationTriple> extractions = segmenter.Extract(parse, tokens);
// note: uses non-coref-canonicalized parse!
Sharpen.Collections.AddAll(extractions, RelationsInFragments(fragments, sentence));
// Set the annotations
sentence.Set(typeof(NaturalLogicAnnotations.EntailedClausesAnnotation), new HashSet <SentenceFragment>(clauses));
sentence.Set(typeof(NaturalLogicAnnotations.EntailedSentencesAnnotation), fragments);
sentence.Set(typeof(NaturalLogicAnnotations.RelationTriplesAnnotation), new List <RelationTriple>(new HashSet <RelationTriple>(extractions)));
// uniq the extractions
if (stripEntailments)
{
sentence.Remove(typeof(NaturalLogicAnnotations.EntailedSentencesAnnotation));
}
}
}
private void CleanupTags(ICoreMap cm, IDictionary <object, bool> cleaned)
{
cm.Remove(typeof(Tags.TagsAnnotation));
foreach (Type key in cm.KeySet())
{
object obj = cm.Get(key);
if (!cleaned.Contains(obj))
{
cleaned[obj] = false;
if (obj is ICoreMap)
{
CleanupTags((ICoreMap)obj, cleaned);
}
else
{
if (obj is ICollection)
{
CleanupTags((ICollection)obj, cleaned);
}
}
cleaned[obj] = true;
}
}
}