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 static void BuildSuggestor()
{
IIO io = new FileIO();
IEnumerable<string> file = io.ReadSourceIterable(testpath + "TIME.ALL");
ITextExtractor it = new BeginMarkerExtraction(file, "*TEXT");
Console.WriteLine("Building suggestion base");
WordSuggestor ws = new WordSuggestor();
ws.addAll(it);
Console.WriteLine("Saving tree");
ITreeIO tio = new TreeIO();
tio.SaveObject(ws, testpath + "WordSuggestions");
}
public static void TestSuggestions()
{
IIO io = new FileIO();
IEnumerable<string> file = io.ReadSourceIterable(testpath + "TIME.ALL");
ITextExtractor it = new BeginMarkerExtraction(file, "*TEXT");
WordSuggestor ws = new WordSuggestor();
ws.AddAllStemmed(it);
var words = ws.Suggestions(.2);
foreach (string s in words) {
Console.WriteLine(s);
}
}
public static void TestWordSuggestions()
{
IIO io = new FileIO();
IEnumerable<string> file = io.ReadSourceIterable(testpath + "TIME.ALL");
ITextExtractor it = new BeginMarkerExtraction(file, "*TEXT");
Console.WriteLine("Building suggestion base");
WordSuggestor ws = new WordSuggestor();
ws.AddAllStemmed(it);
string test = "embassi";
Console.WriteLine("Determining suggestions for '" + test + "':");
var words = ws.WordSuggestions(test, .2);
foreach (string s in words) {
Console.WriteLine(s);
}
}
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);
}
public static void getDocsForQuery(string contentTreeName)
{
Console.SetBufferSize(100, 2000);
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);
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.CreateStemmedDocumentMapTree(tree);
DataTreeBuilder.AddToDataTreeBoyerMoore(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: ");
List<string> retrieved = new List<string>();
foreach (String s in Directory.EnumerateFiles(testpath + @"\datatrees")) {
IDataTree docTree = tio.LoadDataTree(s);
if (queryTree.CompareTo(docTree)) {
Console.Write(" " + docTree.Name);
retrieved.Add(docTree.Name);
totalRetrieved++;
if (expectedArray.Contains(docTree.Name)) {
relevant++;
}
Console.WriteLine("Found---");
Console.WriteLine(docTree.Name);
docTree.PrintDataTree();
Console.WriteLine("---");
}
if (expectedArray.Contains(docTree.Name)) {
Console.WriteLine("Expected---");
Console.WriteLine(docTree.Name);
docTree.PrintDataTree();
Console.WriteLine("---");
}
}
Console.WriteLine();
Console.WriteLine("Precision: " + relevant + "/" + totalRetrieved);
Console.WriteLine("Recall: " + relevant + "/" + (expectedArray.Length));
Console.WriteLine();
Console.WriteLine("---------------------------------");
Thread.Sleep(10000);
IEnumerable<string> fileAll = io.ReadSourceIterable(testpath + "TIME.ALL");
ITextExtractor itAll = new BeginMarkerExtraction(fileAll, "*TEXT");
//int count = 1;
//while (itAll.HasNextContent()) {
// string content = itAll.FindNextContent();
// string name = "" + count;
// if (retrieved.Contains(name) || expectedArray.ToList().Contains(name)) {
// Console.WriteLine("------------------------------------------------------------");
// Console.WriteLine("------------------------------------------------------------");
// Console.WriteLine(name);
// content = Helpers.ConsumeName(content);
// Console.WriteLine(content);
// Console.WriteLine("------------------------------------------------------------");
// Console.WriteLine("------------------------------------------------------------");
// }
// 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;
}
