private void CollectEdges(SolarixGrammarEngineNET.SyntaxTreeNode node, Dictionary <int, int> node2parent, Dictionary <int, string> node2word)
{
node2word.Add(node.GetWordPosition(), node.GetWord());
for (int i = 0; i < node.leafs.Count; ++i)
{
SolarixGrammarEngineNET.SyntaxTreeNode child = node.leafs[i];
node2parent.Add(child.GetWordPosition(), node.GetWordPosition());
CollectEdges(child, node2parent, node2word);
}
}
private void SetLabel(SolarixGrammarEngineNET.SyntaxTreeNode node, string label, bool recursive)
{
if (recursive)
{
List <int> indeces = new List <int>();
CollectSubtreeNodeIndeces(node, indeces);
int k = 0;
foreach (int index in indeces.OrderBy(z => z))
{
if (k == 0)
{
labels[index + 1].Insert(0, label);
}
else
{
labels[index + 1].Insert(0, label);
}
k++;
}
}
else
{
int index = node.GetWordPosition();
labels[index + 1].Insert(0, label);
}
return;
}
public bool ProcessSample(SampleData sample, bool train_sample, bool test_sample)
{
if (wrt_train == null)
{
wrt_train = new System.IO.StreamWriter("syntax_neuro_train.txt");
wrt_test = new System.IO.StreamWriter("syntax_neuro_test.txt");
}
System.IO.StreamWriter wrt = train_sample ? wrt_train : wrt_test;
if (sample.syntax_tree.Count == 3)
{
SolarixGrammarEngineNET.SyntaxTreeNode root = sample.syntax_tree[1];
Dictionary <int, int> node2parent = new Dictionary <int, int>();
Dictionary <int, string> node2word = new Dictionary <int, string>();
node2parent.Add(root.GetWordPosition(), -1);
CollectEdges(root, node2parent, node2word);
foreach (int index in node2word.Select(z => z.Key).OrderBy(z => z))
{
wrt.WriteLine("{0}\t{1}\t{2}", index, node2word[index], node2parent[index]);
}
wrt.WriteLine("");
}
return(true);
}
static void GetChunkNodes(SolarixGrammarEngineNET.SyntaxTreeNode node,
int chunk_index,
Dictionary <int /*word_index*/, int /*chunk_index*/> labels)
{
labels[node.GetWordPosition()] = chunk_index;
for (int i = 0; i < node.leafs.Count; ++i)
{
GetChunkNodes(node.leafs[i], chunk_index, labels);
}
}
private void CollectSubtreeNodeIndeces(SolarixGrammarEngineNET.SyntaxTreeNode node, List <int> word_index)
{
if (node.GetWord() == ".")
{
return;
}
word_index.Add(node.GetWordPosition());
for (int ileaf = 0; ileaf < node.leafs.Count; ++ileaf)
{
SolarixGrammarEngineNET.SyntaxTreeNode leaf = node.leafs[ileaf];
CollectSubtreeNodeIndeces(leaf, word_index);
}
return;
}
void TraverseEdges(SolarixGrammarEngineNET.SyntaxTreeNode token)
{
foreach (SolarixGrammarEngineNET.SyntaxTreeNode leaf in token.leafs)
{
int distance = System.Math.Abs(leaf.GetWordPosition() - token.GetWordPosition());
if (edge_len2count.ContainsKey(distance))
{
edge_len2count[distance] = edge_len2count[distance] + 1;
}
else
{
edge_len2count.Add(distance, 1);
}
TraverseEdges(leaf);
}
}