// static methods
/// <summary>
/// Put the tree in the CoreMap for the sentence, also add any
/// dependency graphs to the sentence, and fill in missing tag annotations.
/// </summary>
/// <remarks>
/// Put the tree in the CoreMap for the sentence, also add any
/// dependency graphs to the sentence, and fill in missing tag annotations.
/// Thread safety note: nothing special is done to ensure the thread
/// safety of the GrammaticalStructureFactory. However, both the
/// EnglishGrammaticalStructureFactory and the
/// ChineseGrammaticalStructureFactory are thread safe.
/// </remarks>
public static void FillInParseAnnotations(bool verbose, bool buildGraphs, IGrammaticalStructureFactory gsf, ICoreMap sentence, IList <Tree> trees, GrammaticalStructure.Extras extras)
{
bool first = true;
foreach (Tree tree in trees)
{
// make sure all tree nodes are CoreLabels
// TODO: why isn't this always true? something fishy is going on
Edu.Stanford.Nlp.Trees.Trees.ConvertToCoreLabels(tree);
// index nodes, i.e., add start and end token positions to all nodes
// this is needed by other annotators down stream, e.g., the NFLAnnotator
tree.IndexSpans(0);
if (first)
{
sentence.Set(typeof(TreeCoreAnnotations.TreeAnnotation), tree);
if (verbose)
{
log.Info("Tree is:");
tree.PennPrint(System.Console.Error);
}
SetMissingTags(sentence, tree);
if (buildGraphs)
{
// generate the dependency graph
// unfortunately, it is necessary to make the
// GrammaticalStructure three times, as the dependency
// conversion changes the given data structure
SemanticGraph deps = SemanticGraphFactory.GenerateCollapsedDependencies(gsf.NewGrammaticalStructure(tree), extras);
SemanticGraph uncollapsedDeps = SemanticGraphFactory.GenerateUncollapsedDependencies(gsf.NewGrammaticalStructure(tree), extras);
SemanticGraph ccDeps = SemanticGraphFactory.GenerateCCProcessedDependencies(gsf.NewGrammaticalStructure(tree), extras);
SemanticGraph enhancedDeps = SemanticGraphFactory.GenerateEnhancedDependencies(gsf.NewGrammaticalStructure(tree));
SemanticGraph enhancedPlusPlusDeps = SemanticGraphFactory.GenerateEnhancedPlusPlusDependencies(gsf.NewGrammaticalStructure(tree));
if (verbose)
{
log.Info("SDs:");
log.Info(deps.ToString(SemanticGraph.OutputFormat.List));
}
sentence.Set(typeof(SemanticGraphCoreAnnotations.CollapsedDependenciesAnnotation), deps);
sentence.Set(typeof(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation), uncollapsedDeps);
sentence.Set(typeof(SemanticGraphCoreAnnotations.CollapsedCCProcessedDependenciesAnnotation), ccDeps);
sentence.Set(typeof(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation), enhancedDeps);
sentence.Set(typeof(SemanticGraphCoreAnnotations.EnhancedPlusPlusDependenciesAnnotation), enhancedPlusPlusDeps);
}
first = false;
}
}
if (trees.Count > 1)
{
sentence.Set(typeof(TreeCoreAnnotations.KBestTreesAnnotation), trees);
}
}
public virtual void TestLemma()
{
SemanticGraph graph = SemanticGraph.ValueOf("[ate subj>Bill dobj>[muffins compound>blueberry]]");
foreach (IndexedWord word in graph.VertexSet())
{
word.SetLemma(word.Word());
}
RunTest("{lemma:ate}", graph, "ate");
Tree tree = Tree.ValueOf("(ROOT (S (NP (PRP I)) (VP (VBP love) (NP (DT the) (NN display))) (. .)))");
graph = SemanticGraphFactory.GenerateCCProcessedDependencies(tree);
foreach (IndexedWord word_1 in graph.VertexSet())
{
word_1.SetLemma(word_1.Word());
}
// This set of three tests also provides some coverage for a
// bizarre error a user found where multiple copies of the same
// IndexedWord were created
RunTest("{}=Obj <dobj {lemma:love}=Pred", graph, "display/NN");
RunTest("{}=Obj <dobj {}=Pred", graph, "display/NN");
RunTest("{lemma:love}=Pred >dobj {}=Obj ", graph, "love/VBP");
}