/// <summary>
/// Gets a copy of this collection
/// </summary>
/// <returns>Copy of this collection</returns>
public override LabeledNodeCollection Copy()
{
// create copy
PropBankLabeledNodeCollection copy = new PropBankLabeledNodeCollection(new PropBankNodeLabel(_label.Type, _label.Feature, _label.Confidence));
// add single nodes
foreach (TreeBankNode singleNode in SingleNodes)
{
copy.AddSingleNode(singleNode);
}
// add split nodes
foreach (List <TreeBankNode> splitNode in SplitNodes)
{
copy.AddSplitNode(splitNode);
}
return(copy);
}
/// <summary>
/// Sets the label on this node
/// </summary>
/// <param name="label">Label to use</param>
/// <param name="syncWithRootNodeCollection">Whether or not to add this node to the corresponding node collection on the root. The
/// root collections are shortcut collections that allow quick searching for particular node types. Thus, the collections need to
/// remain synchronized with the labels that are applied to the nodes. Passing true here will perform this synchronization. If you
/// will do the synchronization on your own later, pass false.</param>
public void SetLabel(PropBankNodeLabel label, bool syncWithRootNodeCollection)
{
// first remove current node from previous node collection if it is present
if (_label != null)
{
PropBankLabeledNodeCollection nodes = GetLabeledNodeCollection(label, false);
if (nodes != null && nodes.ContainsSingleNode(this))
{
nodes.RemoveSingleNode(this);
}
}
_label = label;
// add node to root's collection if needed
if (syncWithRootNodeCollection)
{
GetLabeledNodeCollection(_label, true).AddSingleNode(this);
}
}
/// <summary>
/// Gets labeled node collection
/// </summary>
/// <param name="label">Label of node collection to get</param>
/// <param name="createIfMissing">Whether or not to create and return a new node collection if none exists</param>
/// <returns>Labeled node collection</returns>
public PropBankLabeledNodeCollection GetLabeledNodeCollection(PropBankNodeLabel label, bool createIfMissing)
{
// look for an existing collection with the given label
foreach (PropBankLabeledNodeCollection nodeCollection in LabeledNodeCollections)
{
if (nodeCollection.Label == label)
{
return(nodeCollection);
}
}
if (!createIfMissing)
{
return(null);
}
// return new collection
PropBankLabeledNodeCollection newCollection = new PropBankLabeledNodeCollection(label);
LabeledNodeCollections.Add(newCollection);
return(newCollection);
}
/// <summary>
/// Gets a predicate tree for a PropBank propositions entry
/// </summary>
/// <param name="vi">VerbInfo specifying tree to look up</param>
/// <returns>PropBankNode</returns>
public PropBankNode GetPropBankTree(VerbInfo vi)
{
TreeBankNode parse = GetParseTree(vi.File, vi.SentenceNumber);
PropBankNode predTree = new PropBankNode(parse);
predTree.Information = vi;
// label information is space-delimited
string[] labels = vi.LabeledNodeLocations.Split(' ');
foreach (string label in labels)
{
// label columns are dash-delimited
string[] labelCols = label.Split('-');
// get label type
PropBankNodeLabel.NodeType labelType = PropBankNodeLabel.GetNodeType(labelCols[1]);
// get label feature if any
PropBankNodeLabel.NodeFeature labelFeature = PropBankNodeLabel.NodeFeature.None;
if (labelCols.Length > 2)
{
// sometimes the feature is the actual preposition, so this might fail
string featureStr = labelCols[2];
if (!PropBankNodeLabel.TryGetNodeFeature(featureStr, out labelFeature))
{
// use PRP as the feature, which we have added for this case
featureStr = "PRP";
labelFeature = PropBankNodeLabel.GetNodeFeature(featureStr);
}
if (labelCols.Length > 3)
{
throw new Exception("Missed feature");
}
}
// create new labeled node collection
PropBankLabeledNodeCollection labeledNodes = new PropBankLabeledNodeCollection(new PropBankNodeLabel(labelType, labelFeature, 1));
AddNodesToCollection(predTree, labelCols[0], labeledNodes);
// add to root's list of nodes
predTree.LabeledNodeCollections.Add(labeledNodes);
}
// make sure one of the predicate leaves has the leaf number from the propositions file entry
bool foundMatch = false;
foreach (PropBankNode predicateNode in predTree.PredicateNodes)
{
foreach (PropBankNode leaf in predicateNode.Leaves)
{
if (leaf.LeafNumber == vi.LeafNumber)
{
foundMatch = true;
break;
}
}
if (foundMatch)
{
break;
}
}
if (!foundMatch)
{
throw new Exception("Mismatch between VerbInfo predicate leaf number and actual predicate leaf number");
}
return(predTree);
}
/// <summary>
/// Marks argument nodes from the current node in the corresponding parse from a different TreeBank. This is used when
/// transferring PropBank annotations to parse trees other than those distributed in the TreeBank (e.g., those produced
/// by an automatic syntactic parser).
/// </summary>
/// <param name="treeBankEngine">Initialized TreeBank engine from which to pull the parse tree to mark PropBank arguments within</param>
/// <returns>PropBank node, or null if all arguments couldn't be minimally transferred to the other parse tree. An argument
/// is minimally transferred if the corresponding node in the other parse tree subsumes precisely the same text as the node in the
/// current parse tree. Sometimes this is not possible due to parse errors.</returns>
public PropBankNode MarkArgumentNodesIn(TreeBankEngine treeBankEngine)
{
if (!IsRoot)
{
throw new Exception("Attempted to transform non-root node");
}
// get mrg file in other tree bank
string treeBankMrgFile = treeBankEngine.GetFullMrgPath(MrgFile.Substring(MrgFile.LastIndexOf(Path.DirectorySeparatorChar) + 1));
// need a PropBank root to mark arguments within
PropBankNode pbRoot = new PropBankNode(treeBankEngine.GetParseTree(treeBankMrgFile, SentenceNumber));
// make sure we got the right sentence
if (pbRoot.SurfaceText != SurfaceText)
{
throw new Exception("Failed to convert root to Charniak-parsed version");
}
// Add information to root. Ignore leaf number and argument info for now - we'll set them at the end.
treeBankMrgFile = treeBankMrgFile.Substring(treeBankEngine.MrgPath.Length);
VerbInfo pbInfo = Information;
pbRoot.Information = new VerbInfo(pbInfo.Verb, treeBankMrgFile, pbInfo.SentenceNumber,
-1, pbInfo.Tagger, pbInfo.RoleSetId,
pbInfo.VForm, pbInfo.VTense, pbInfo.VAspect,
pbInfo.VPerson, pbInfo.VVoice, "");
// transfer all argument node lists
foreach (PropBankLabeledNodeCollection nodeCollection in LabeledNodeCollections)
{
// new node collection
PropBankLabeledNodeCollection otherNodeCollection = new PropBankLabeledNodeCollection(new PropBankNodeLabel(nodeCollection.Label.Type, nodeCollection.Label.Feature, nodeCollection.Label.Confidence));
// get single nodes
foreach (PropBankNode singleNode in nodeCollection.SingleNodes)
{
if (!singleNode.IsNullElement)
{
// get argument node from other parse tree
PropBankNode otherArgNode = (PropBankNode)pbRoot.GetMinimallySubsumingNode(singleNode.FirstToken, singleNode.LastToken);
if (otherArgNode == null)
{
return(null);
}
otherNodeCollection.AddSingleNode(otherArgNode);
}
}
// get split arguments
foreach (List <TreeBankNode> splitArg in nodeCollection.SplitNodes)
{
List <TreeBankNode> otherSplitArg = new List <TreeBankNode>();
// get each node in the split argument
foreach (PropBankNode splitArgNode in splitArg)
{
if (!splitArgNode.IsNullElement)
{
// get split node in other tree
PropBankNode otherSplitArgNode = (PropBankNode)pbRoot.GetMinimallySubsumingNode(splitArgNode.FirstToken, splitArgNode.LastToken);
if (otherSplitArgNode == null)
{
return(null);
}
otherSplitArg.Add(otherSplitArgNode);
}
}
// if only one node of the split arg was non-null, at that node as a single
if (otherSplitArg.Count == 1)
{
otherNodeCollection.AddSingleNode(otherSplitArg.First());
}
// otherwise, add the split arg normally
else if (otherSplitArg.Count >= 2)
{
otherNodeCollection.AddSplitNode(otherSplitArg);
}
}
// add coref list if we found non-null nodes
if (otherNodeCollection.SingleNodes.Count > 0 || otherNodeCollection.SplitNodes.Count > 0)
{
pbRoot.LabeledNodeCollections.Add(otherNodeCollection);
}
}
// return null if we didn't find any argument node lists with non-null nodes
if (pbRoot.LabeledNodeCollections.Count == 0)
{
return(null);
}
// set leaf number and argument information
pbRoot.Information.LeafNumber = pbRoot.PredicateNodes.First().Leaves[0].LeafNumber;
pbRoot.Information.LabeledNodeLocations = pbRoot.LabeledNodeLocations;
return(pbRoot);
}