public void DocumentMappingTest2()
{
IBaseTree tree = setUpTree();
IDataTree dataTree = DataTreeBuilder.CreateDocumentMappedTree(tree, toMap);
//make sure the root has the correct number of children
Assert.AreEqual(dataTree.Root.Children.Count, 2);
//make sure each of the children has the correct weight
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "tree").FirstOrDefault().Weight, 3);
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "others").FirstOrDefault().Weight, 2);
//make sure a branch on the content tree which is not existent in the mapped tree does not exist
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "nonexistent").FirstOrDefault(), null);
//make sure each child's connections have their own correct weight
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "tree").FirstOrDefault().EndPoint.Children.Where(x => x.EndPoint.Keyword == "content").FirstOrDefault().Weight, 1);
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "tree").FirstOrDefault().EndPoint.Children.Where(x => x.EndPoint.Keyword == "mapped").FirstOrDefault().Weight, 1);
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "others").FirstOrDefault().EndPoint.Children.Where(x => x.EndPoint.Keyword == "series").FirstOrDefault().Weight, 1);
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "others").FirstOrDefault().EndPoint.Children.Where(x => x.EndPoint.Keyword == "going").FirstOrDefault().Weight, 1);
//make sure children connections do not cross
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "others").FirstOrDefault().EndPoint.Children.Where(x => x.EndPoint.Keyword == "content").FirstOrDefault(), null);
//make sure leaf children do not have any connections
Assert.AreEqual(dataTree.Root.Children.Where(c => c.EndPoint.Keyword == "tree").FirstOrDefault().EndPoint.Children.Where(x => x.EndPoint.Keyword == "content").FirstOrDefault().EndPoint.Children.Count, 0);
}
public static void MakeTrees(string contentTreeName)
{
IIO io = new FileIO();
IEnumerable<string> file = io.ReadSourceIterable(testpath + "TIME.ALL");
ITextExtractor it = new BeginMarkerExtraction(file, "*TEXT");
ITreeIO tio = new TreeIO();
IBaseTree tree = tio.LoadBaseTree(testpath + contentTreeName);
int count = 1;
while (it.HasNextContent()) {
string content = it.FindNextContent();
//Console.WriteLine("-----");
string name = "" + count;
//Console.WriteLine(name);
content = Helpers.ConsumeName(content);
//Console.WriteLine(content);
IDataTree datatree = DataTreeBuilder.CreateDocumentMappedTree(tree);
//Console.WriteLine(tree);
//Console.WriteLine(datatree.GetBaseTree());
DataTreeBuilder.AddToDataTree(datatree, content);
datatree.Name = name;
tio.SaveDataTree(datatree, testpath + @"\datatrees\" + name + ".dtree");
//Console.WriteLine(datatree.MappedWords);
count++;
}
}
public IDataTree setUpDataTree(IBaseTree baseTree)
{
IDataTree data = DataTreeBuilder.CreateDocumentMappedTree(baseTree);
DataTreeBuilder.AddToDataTree(data, toMap);
return(data);
}
public void DocumentMapNonexistentContentEnumberable()
{
IBaseTree tree = setUpTree();
IDataTree dataTree = DataTreeBuilder.CreateDocumentMappedTree(tree);
DataTreeBuilder.AddAllToDataTree(dataTree, null);
}
public void DocumentMapChangeBaseTree()
{
IBaseTree tree = setUpTree();
IDataTree dataTree = DataTreeBuilder.CreateDocumentMappedTree(tree);
DataTreeBuilder.AddAllToDataTree(dataTree, toMapEnumerable);
dataTree.SetBaseTree(null);
Assert.AreEqual(dataTree.GetBaseTree(), null);
}
public void CompareTrees()
{
IBaseTree tree = setUpTree();
IDataTree originalDataTree = DataTreeBuilder.CreateDocumentMappedTree(tree);
DataTreeBuilder.AddToDataTree(originalDataTree, toMap);
IDataTree probablyMatches = DataTreeBuilder.CreateDocumentMappedTree(tree);
DataTreeBuilder.AddToDataTree(probablyMatches, shouldProbablyMatch);
IDataTree notAMatch = DataTreeBuilder.CreateDocumentMappedTree(tree);
DataTreeBuilder.AddToDataTree(notAMatch, shouldProbablyNotMatch);
Assert.IsTrue(originalDataTree.CompareTo(originalDataTree));
Assert.IsTrue(originalDataTree.CompareTo(probablyMatches));
Assert.IsFalse(originalDataTree.CompareTo(notAMatch));
}
private void queryOkay_Click(object sender, RoutedEventArgs e)
{
string query = queryBox.Text;
List <string> matches = new List <string>();
using (Ookii.Dialogs.Wpf.ProgressDialog dial = new ProgressDialog()) {
IDataTree queryTree = DataTreeBuilder.CreateDocumentMappedTree(baseTree);
DataTreeBuilder.AddToDataTreeBoyerMoore(queryTree, query);
ITreeIO tio = new TreeIO();
foreach (String s in Directory.EnumerateFiles(documentLabel.Content.ToString()))
{
IDataTree docTree = tio.LoadDataTree(s);
if (queryTree.CompareTo(docTree))
{
matches.Add(s);
}
}
}
if (matches.Count > 0)
{
StringBuilder sb = new StringBuilder("Documents which match \"" + query + "\":");
foreach (string s in matches)
{
string s2 = System.IO.Path.GetFileName(s);
s2 = new string(s2.Take(s2.LastIndexOf('.')).ToArray());
documents.Add(s2);
sb.Append(" " + s2);
}
MessageBox.Show(sb.ToString());
}
else
{
MessageBox.Show("No documents match the query \"" + query + "\"");
}
queryPopup.IsOpen = false;
}
public void DocumentMapNonexistentContent2()
{
IBaseTree tree = setUpTree();
IDataTree dataTree = DataTreeBuilder.CreateDocumentMappedTree(tree, null);
}
public void DocumentMapNonexistentBaseTree2()
{
IDataTree dataTree = DataTreeBuilder.CreateDocumentMappedTree(null, toMap);
}
public static void Compare(string contentTreeName)
{
Console.SetBufferSize(100, 20000);
IIO io = new FileIO();
IEnumerable<string> file = io.ReadSourceIterable(testpath + "TIME.QUE");
IEnumerable<string> expectedResults = io.ReadSourceIterable(testpath + "TIME.REL");
var resultsEnum = expectedResults.GetEnumerator();
ITextExtractor it = new BeginMarkerExtraction(file, "*FIND");
ITreeIO tio = new TreeIO();
IBaseTree tree = tio.LoadBaseTree(testpath + contentTreeName);
double totalRecall = 0;
double totalPrecision = 0;
double bestRecall = -1;
double worstRecall = 2;
double bestPrecision = -1;
double worstPrecision = 2;
double totalGoodPrecision = 0;
double count = 0;
while (it.HasNextContent()) {
string query = it.FindNextContent();
Console.WriteLine("---------------------------------");
string queryName = Helpers.GetNameWhenFirst(query);
Console.WriteLine("Query: " + queryName);
query = Helpers.ConsumeName(query);
Console.WriteLine(query);
IDataTree queryTree = DataTreeBuilder.CreateDocumentMappedTree(tree);
DataTreeBuilder.AddToDataTree(queryTree, query);
queryTree.PrintDataTree();
Console.WriteLine("Expected Results: ");
while(string.IsNullOrEmpty(resultsEnum.Current))
resultsEnum.MoveNext();
string expected = Helpers.ConsumeName(resultsEnum.Current);
Console.WriteLine(expected);
resultsEnum.MoveNext();
expected = expected.Trim();
string[] expectedArray = expected.Split(' ');
double relevant = 0;
double totalRetrieved = 0;
Console.WriteLine("Actual Results: ");
foreach(String s in Directory.EnumerateFiles(testpath + @"\datatrees")) {
IDataTree docTree = tio.LoadDataTree(s);
if (queryTree.CompareTo(docTree)) {
Console.WriteLine(" Found: " + docTree.Name);
//docTree.PrintDataTree();
totalRetrieved++;
if (expectedArray.Contains(docTree.Name)) {
relevant++;
}
} else if (expectedArray.Contains(docTree.Name)) {
Console.WriteLine(" Expected: " + docTree.Name);
//docTree.PrintDataTree();
}
}
Console.WriteLine();
Console.WriteLine("Precision: " + relevant + "/" + totalRetrieved );
Console.WriteLine("Recall: " + relevant + "/" + (expectedArray.Length));
Console.WriteLine();
//Console.ReadLine();
count++;
double recall = relevant / expectedArray.Length;
double precision = 0;
if (totalRetrieved > 0) {
precision = relevant / totalRetrieved;
}
totalPrecision += precision;
totalRecall += recall;
if (precision > bestPrecision) {
bestPrecision = precision;
}
if (precision < worstPrecision) {
worstPrecision = precision;
}
if(recall > bestRecall) {
bestRecall = recall;
}
if(recall < worstRecall) {
worstRecall = recall;
}
if (recall > .5) {
totalGoodPrecision += precision;
}
}
Console.WriteLine("-------------------");
Console.WriteLine("Average Precision: " + totalPrecision / count);
Console.WriteLine("Average Recall: " + totalRecall / count);
Console.WriteLine("Worst Precision: " + worstPrecision);
Console.WriteLine("Worst Recall: " + worstRecall);
Console.WriteLine("Best Precision: " + bestPrecision);
Console.WriteLine("Best Recall: " + bestRecall);
//Console.WriteLine("Average Good Recall Precision: " + totalGoodPrecision / count);
}
private void formatOkay_Click(object sender, RoutedEventArgs e)
{
if (baseTree == null)
{
MessageBox.Show("Please select a content tree for the data tree.");
return;
}
if (formatBox.SelectedIndex == -1)
{
formatBox.BorderBrush = Brushes.Red;
return;
}
if (string.IsNullOrEmpty(documentFormatBox.Text))
{
documentFormatBox.BorderBrush = Brushes.Red;
return;
}
OpenFileDialog ofd = new OpenFileDialog();
ofd.FileName = "Tree";
ofd.DefaultExt = ".txt";
Nullable <bool> result = ofd.ShowDialog();
if (result == true)
{
string filename = ofd.FileName;
documentLabel.Content = filename + "datatrees";
using (Ookii.Dialogs.Wpf.ProgressDialog dial = new ProgressDialog()) {
dial.ProgressBarStyle = ProgressBarStyle.MarqueeProgressBar;
dial.Show();
dial.Description = "Analyzing text...";
IIO io = new FileIO();
ITextExtractor it = null;
switch (formatBox.SelectedIndex)
{
case 0:
string text = io.ReadSource(filename);
it = new XMLTextExtractor(text, documentFormatBox.Text);
break;
case 1:
var texts = io.ReadSourceIterable(filename);
it = new BeginMarkerExtraction(texts, documentFormatBox.Text);
break;
default:
throw new InvalidOperationException();
}
documents = new ObservableCollection <string>();
while (it.HasNextContent())
{
string content = it.FindNextContent();
string name = Helpers.GetNameWhenFirst(content);
documents.Add(name);
IDataTree tree = DataTreeBuilder.CreateDocumentMappedTree(baseTree);
DataTreeBuilder.AddToDataTree(tree, content);
ITreeIO tio = new TreeIO();
tio.SaveDataTree(tree, filename + @"datatrees\" + name + ".dtree");
}
documentList.ItemsSource = documents;
}
}
buildDataTreePopup.IsOpen = false;
}