public void TestAccuracy()
{
int correct = 0, total = 0;
var mostFrequentTreeAutoSemantic = new MostFrequentTreeAutoSemantic(wordNet, fsm);
var treeBank1 = new TreeBankDrawable("../../../new-trees/");
var treeBank2 = new TreeBankDrawable("../../../old-trees/");
for (var i = 0; i < treeBank1.Size(); i++)
{
var parseTree1 = treeBank1.Get(i);
var parseTree2 = treeBank2.Get(i);
mostFrequentTreeAutoSemantic.AutoSemantic(parseTree1);
var nodeDrawableCollector1 =
new NodeDrawableCollector((ParseNodeDrawable)parseTree1.GetRoot(), new IsTurkishLeafNode());
var leafList1 = nodeDrawableCollector1.Collect();
var nodeDrawableCollector2 =
new NodeDrawableCollector((ParseNodeDrawable)parseTree2.GetRoot(), new IsTurkishLeafNode());
var leafList2 = nodeDrawableCollector2.Collect();
for (var j = 0; j < leafList1.Count; j++)
{
total++;
var parseNode1 = leafList1[j];
var parseNode2 = leafList2[j];
if (parseNode1.GetLayerData(ViewLayerType.SEMANTICS) != null && parseNode1
.GetLayerData(ViewLayerType.SEMANTICS).Equals(parseNode2.GetLayerData(ViewLayerType.SEMANTICS)))
{
correct++;
}
}
}
Assert.AreEqual(475, total);
Assert.AreEqual(260, correct);
}
/**
* <summary> Constructor for the DataSetGenerator which takes input the data directory, the pattern for the training files
* included, includePunctuation, and an instanceGenerator. The constructor loads the treeBank from the given directory
* including the given files having the given pattern. If punctuations are not included, they are removed from
* the data.</summary>
* <param name="directory">Directory where the treeBank files reside.</param>
* <param name="pattern">Pattern of the tree files to be included in the treeBank. Use "." for all files.</param>
* <param name="includePunctuation">If true, punctuation symbols are also included in the dataset, false otherwise.</param>
* <param name="instanceGenerator">The instance generator used to generate the dataSet.</param>
*/
public DataSetGenerator(string directory, string pattern, bool includePunctuation,
InstanceGenerator.InstanceGenerator instanceGenerator)
{
_treeBank = new TreeBankDrawable(directory, pattern);
this.instanceGenerator = instanceGenerator;
if (!includePunctuation)
{
_treeBank.StripPunctuation();
}
}
/**
* <summary> Constructor for the DisambiguationCorpusGenerator which takes input the data directory and the pattern for the
* training files included. The constructor loads the treebank from the given directory including the given files
* having the given pattern.</summary>
*
* <param name="directory">Directory where the treebank files reside.</param>
* <param name="pattern">Pattern of the tree files to be included in the treebank. Use "." for all files.</param>
*/
public TreeDisambiguationCorpusGenerator(string directory, string pattern)
{
_treeBank = new TreeBankDrawable(directory, pattern);
}
public ConstituencyToDependencyTreeBank(TreeBankDrawable treeBank)
{
_treeBank = treeBank;
}
/**
* <summary> Constructor for the NERCorpusGenerator which takes input the data directory and the pattern for the training files
* included. The constructor loads the treebank from the given directory including the given files having the given
* pattern.</summary>
*
* <param name="directory">Directory where the treebank files reside.</param>
* <param name="pattern">Pattern of the tree files to be included in the treebank. Use "." for all files.</param>
*/
public NERCorpusGenerator(string directory, string pattern)
{
_treeBank = new TreeBankDrawable(directory, pattern);
}
