private List <Decision> SetToAndAddUniversalDependency(int startIndex, int headIndex, List <WordNodePair> wordNodePairList, int finishIndex, ParseNodeDrawable parent)
{
var decisions = new List <Decision>();
for (var i = startIndex; i <= finishIndex; i++)
{
if (i != headIndex)
{
var parentData = parent.GetData().GetName();
var firstChild = parent.GetChild(0).GetData().GetName();
string secondChild = null, thirdChild = null;
if (parent.NumberOfChildren() > 1)
{
secondChild = parent.GetChild(1).GetData().GetName();
}
if (parent.NumberOfChildren() > 2)
{
thirdChild = parent.GetChild(2).GetData().GetName();
}
if (parent.NumberOfChildren() == 2 && parentData.Equals("S") && firstChild.Equals("NP"))
{
decisions.Add(new Decision(startIndex + decisions.Count, headIndex - i, "NSUBJ"));
}
else if (parent.NumberOfChildren() == 3 && parentData.Equals("S") && firstChild.Equals("NP") && secondChild.Equals("VP") && Word.IsPunctuation(thirdChild))
{
if (!wordNodePairList[i].GetWord().IsPunctuation())
{
decisions.Add(new Decision(startIndex + decisions.Count, headIndex - i, "NSUBJ"));
}
else
{
decisions.Add(new Decision(startIndex + decisions.Count, headIndex - i, "PUNCT"));
}
}
else
{
var dependent = wordNodePairList[i].GetNode().GetData().GetName();
var head = wordNodePairList[headIndex].GetNode().GetData().GetName();
var condition1 = wordNodePairList[i].GetNode().GetData().IsPunctuation();
var condition2 = wordNodePairList[headIndex].GetNode().GetData().IsPunctuation();
decisions.Add(new Decision(startIndex + decisions.Count, headIndex - i, FindData(dependent, head, condition1, condition2, wordNodePairList[i].GetWord(), wordNodePairList[headIndex].GetWord())));
}
}
else
{
decisions.Add(new Decision(-1, 0, null));
}
}
return(decisions);
}
private void NodeModify(ParseNodeDrawable parseNode)
{
_nodeModifier.Modifier(parseNode);
for (var i = 0; i < parseNode.NumberOfChildren(); i++)
{
NodeModify((ParseNodeDrawable)parseNode.GetChild(i));
}
}
protected bool IsLastWordOfVerbAsNoun(ParseNodeDrawable parseNode)
{
ParseNodeDrawable parent = (ParseNodeDrawable)parseNode.GetParent();
ParseNodeDrawable grandParent = (ParseNodeDrawable)parent.GetParent();
if (parent.IsLastChild(parseNode) && parent.GetData().GetName().Equals("VP"))
{
if (grandParent != null && grandParent.GetData().GetName().Equals("NP") &&
grandParent.GetChild(0).Equals(parent) &&
grandParent.NumberOfChildren() == 2 && grandParent.GetChild(1).GetData().GetName().Equals("NP"))
{
return(true);
}
}
return(false);
}
private List <ParseNodeDrawable> FindNodes(ParseNodeDrawable node)
{
var list = new List <ParseNodeDrawable>();
for (var i = 0; i < node.NumberOfChildren(); i++)
{
var child = (ParseNodeDrawable)node.GetChild(i);
if (node.GetLayerInfo() == null)
{
if (node.NumberOfChildren() == 1 && ((ParseNodeDrawable)node.GetChild(0)).GetLayerInfo() == null)
{
list.Add(node);
}
list.AddRange(FindNodes(child));
}
}
return(list);
}
private void CollectNodes(ParseNodeDrawable parseNode, List <ParseNodeDrawable> collected)
{
if (_condition == null || _condition.Satisfies(parseNode))
{
collected.Add(parseNode);
}
for (var i = 0; i < parseNode.NumberOfChildren(); i++)
{
CollectNodes((ParseNodeDrawable)parseNode.GetChild(i), collected);
}
}
private bool ContainsChild(ParseNodeDrawable parent, ParseNodeDrawable child)
{
for (var i = 0; i < parent.NumberOfChildren(); i++)
{
if (GetParent((ParseNodeDrawable)parent.GetChild(i)).Equals(GetParent(child)))
{
return(true);
}
}
return(false);
}
public List <Decision> MakeDecisions(int firstIndex, int lastIndex, List <WordNodePair> wordNodePairList, ParseNodeDrawable node, List <TreeEnsembleModel> models)
{
if (node.NumberOfChildren() == 3 && node.GetChild(1).GetData().GetName().Equals("CONJP"))
{
var decisions = new List <Decision>();
decisions.Add(new Decision(-1, 0, null));
decisions.Add(new Decision((lastIndex + firstIndex) / 2, lastIndex - ((lastIndex + firstIndex) / 2), "CC"));
decisions.Add(new Decision(lastIndex, firstIndex - lastIndex, "CONJ"));
return(decisions);
}
var headIndex = FindHeadIndex(firstIndex, lastIndex, wordNodePairList);
return(SetToAndAddUniversalDependency(firstIndex, headIndex, wordNodePairList, lastIndex, node));
}
private string ConvertToString(ParseNodeDrawable parseNode)
{
if (parseNode.IsLeaf())
{
return(_converter.LeafConverter(parseNode));
}
var st = "";
for (var i = 0; i < parseNode.NumberOfChildren(); i++)
{
st += ConvertToString((ParseNodeDrawable)parseNode.GetChild(i));
}
return(st);
}
public List <Decision> MakeDecisions(int firstIndex, int lastIndex, List <WordNodePair> wordNodePairList, ParseNodeDrawable node, List <TreeEnsembleModel> models)
{
var testData = new List <Attribute.Attribute>(lastIndex + 1 - firstIndex);
string classInfo;
var list = new List <Tuple <int, int> >();
var decisions = new List <Decision>();
for (var i = 0; i < lastIndex + 1 - firstIndex; i++)
{
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().GetPos()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().GetRootPos()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().ContainsTag(MorphologicalTag.ABLATIVE).ToString()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().ContainsTag(MorphologicalTag.DATIVE).ToString()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().ContainsTag(MorphologicalTag.GENITIVE).ToString()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().ContainsTag(MorphologicalTag.NOMINATIVE).ToString()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().ContainsTag(MorphologicalTag.ACCUSATIVE).ToString()));
testData.Add(new Attribute.DiscreteAttribute(wordNodePairList[firstIndex + i].GetWord().GetParse().ContainsTag(MorphologicalTag.PROPERNOUN).ToString()));
}
switch (lastIndex + 1 - firstIndex)
{
case 2:
classInfo = models[1].Predict(new Instance("", testData));
list = FindList(2, classInfo);
break;
case 3:
classInfo = models[2].Predict(new Instance("", testData));
list = FindList(3, classInfo);
break;
case 4:
classInfo = models[3].Predict(new Instance("", testData));
list = FindList(4, classInfo);
break;
case 5:
classInfo = models[4].Predict(new Instance("", testData));
list = FindList(5, classInfo);
break;
case 6:
classInfo = models[5].Predict(new Instance("", testData));
list = FindList(6, classInfo);
break;
case 7:
classInfo = models[6].Predict(new Instance("", testData));
list = FindList(7, classInfo);
break;
default:
break;
}
var headIndex = FindHeadIndex(list, firstIndex, lastIndex);
for (var i = 0; i < list.Count; i++)
{
var fromWord = wordNodePairList[firstIndex + list[i].Item1].GetWord();
var toWord = wordNodePairList[firstIndex + list[i].Item2].GetWord();
var headWord = wordNodePairList[headIndex].GetWord();
var attributes = new List <Attribute.Attribute>(29);
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().GetPos()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().GetRootPos()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().ContainsTag(MorphologicalTag.ABLATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().ContainsTag(MorphologicalTag.DATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().ContainsTag(MorphologicalTag.GENITIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().ContainsTag(MorphologicalTag.NOMINATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().ContainsTag(MorphologicalTag.ACCUSATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetParse().ContainsTag(MorphologicalTag.PROPERNOUN).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().GetPos()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().GetRootPos()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().ContainsTag(MorphologicalTag.ABLATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().ContainsTag(MorphologicalTag.DATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().ContainsTag(MorphologicalTag.GENITIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().ContainsTag(MorphologicalTag.NOMINATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().ContainsTag(MorphologicalTag.ACCUSATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(toWord.GetParse().ContainsTag(MorphologicalTag.PROPERNOUN).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().GetPos()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().GetRootPos()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().ContainsTag(MorphologicalTag.ABLATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().ContainsTag(MorphologicalTag.DATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().ContainsTag(MorphologicalTag.GENITIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().ContainsTag(MorphologicalTag.NOMINATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().ContainsTag(MorphologicalTag.ACCUSATIVE).ToString()));
attributes.Add(new Attribute.DiscreteAttribute(headWord.GetParse().ContainsTag(MorphologicalTag.PROPERNOUN).ToString()));
if (fromWord.GetSemantic() == null || headWord.GetSemantic() == null)
{
attributes.Add(new Attribute.DiscreteAttribute("null"));
}
else
{
attributes.Add(new Attribute.DiscreteAttribute(fromWord.GetSemantic().Equals(headWord.GetSemantic()).ToString()));
}
attributes.Add(new Attribute.DiscreteAttribute(node.GetData().GetName()));
var firstChild = "null";
var secondChild = "null";
var thirdChild = "null";
if (node.NumberOfChildren() > 0)
{
firstChild = node.GetChild(0).GetData().GetName();
}
if (node.NumberOfChildren() > 1)
{
secondChild = node.GetChild(1).GetData().GetName();
}
if (node.NumberOfChildren() > 2)
{
thirdChild = node.GetChild(2).GetData().GetName();
}
attributes.Add(new Attribute.DiscreteAttribute(firstChild));
attributes.Add(new Attribute.DiscreteAttribute(secondChild));
attributes.Add(new Attribute.DiscreteAttribute(thirdChild));
decisions.Add(new Decision(firstIndex + list[i].Item1, list[i].Item2 - list[i].Item1, models[0].Predict(new Instance("", attributes))));
}
AddHeadToDecisions(decisions, headIndex);
return(decisions);
}