private void getMatrixButton_Click(object sender, RoutedEventArgs e)
{
DocCollection.calcMatrixForDisplay();
buildandBindTable(dataGrid1, DocCollection.termList, DocCollection.weightMatrix, "Filename\\Term");
dataGrid1.Visibility = Visibility.Visible;
ProgressLabel.Content = "Default data set loaded.";
}
private void clearColButton_Click(object sender, RoutedEventArgs e)
{
DocCollection.cleanCollection();
dataGrid1.Visibility = Visibility.Collapsed;
docWriter.deleteAllFiles(filepathGen.Text);
ProgressLabel.Content = "Data collection cleared";
}
/// <summary>
/// takes in a query and a document and returns it's cosine similiarity
/// </summary>
/// <param name="query"></param>
/// <param name="doc"></param>
/// <returns></returns>
static public double cosineWeight(Query query, Doc doc)
{
double term1 = 0;
double bTerm2 = 0;
double bTerm3 = 0;
if (query == null || query.qTerms == null || query.qTerms.Count == 0)
{
return(0);
}
foreach (var entry in doc.dict())
{
bTerm2 += Math.Pow((DocCollection.calcIDF(entry.Key) * entry.Value), 2);//document's weights squared
}
bTerm2 = Math.Sqrt(bTerm2);
foreach (var queryTerm in query.qTerms)
{
if (doc.dict().ContainsKey(queryTerm.term))
{
term1 += queryTerm.weight * //query's weights
((double)doc.dict()[queryTerm.term] * DocCollection.calcIDF(queryTerm.term)); //TFIDF weight for term in collection & doc
}
bTerm3 += Math.Pow(queryTerm.weight, 2); //query's weights squared
}
bTerm3 = Math.Sqrt(bTerm3);
return(term1 / (bTerm2 * bTerm3));//cosine value
}
private void openFile_Click(object sender, RoutedEventArgs e)
{
Value row;
if (dataGrid.SelectedIndex > -1)
{
row = dataGrid.SelectedItem as Value;
var fName = row.term;
DocCollection.openDoc(fName);
}
}
private void removeDoc_Click(object sender, RoutedEventArgs e)
{
var str = testDocLoadtextBox.Text;
if (DocCollection.findDoc(str) != null)
{
DocCollection.removeDoc(str);
docLoadLabel.Content = "None";
docLoadTermsLabel.Content = "0";
dataGrid2.ItemsSource = null;
dataGrid2.Items.Refresh();
}
}
private void button_Click(object sender, RoutedEventArgs e)
{
Query newQ = new Query();
var name = testDocLoadtextBox.Text;
if (DocCollection.findDoc(name) == null)
{
testCVLabel.Content = "Document not\nloaded for testing.";
}
else if (newQ.setQuery(testQTtextBox.Text, testQWtextBox.Text))
{
testCVLabel.Content = Math.Round(Similarity.cosineWeight(newQ, DocCollection.findDoc(name)), 6);
}
}
private void loadDocTest_Click(object sender, RoutedEventArgs e)
{
var str = testDocLoadtextBox.Text;
List <string> headers;
List <string> freq;
List <string> idf;
List <string> weight;
List <List <string> > freqCon;
if (DocCollection.findDoc(str) != null)
{
headers = new List <string>();
freq = new List <string>();
freqCon = new List <List <string> >();
idf = new List <string>();
weight = new List <string>();
foreach (var term in DocCollection.findDoc(str).dict())
{
headers.Add(term.Key);
freq.Add(term.Value.ToString());
idf.Add(Math.Round(DocCollection.calcIDF(term.Key), 4).ToString());
weight.Add(Math.Round((DocCollection.calcIDF(term.Key) * term.Value), 4).ToString());
}
freq.Insert(0, "Freq");
idf.Insert(0, "IDF");
weight.Insert(0, "Weight");
freqCon.Add(freq);
freqCon.Add(idf);
freqCon.Add(weight);
buildandBindTable(dataGrid2, headers, freqCon, "Term");
docLoadLabel.Content = DocCollection.findDoc(str).getFullName();
docLoadTermsLabel.Content = DocCollection.findDoc(str).dict().Count;
}
}
/// <summary>
/// reads in the document and passes each word to the stemmer
/// </summary>
/// <param name="fileName"></param>
public static void parseDoc(string fileName)
{
if (DocCollection.containsDoc(fileName))
{
return;
}
Doc currentDoc = new Doc(fileName);
string line;
try
{
using (TextReader input = File.OpenText(fileName))
{
while ((line = input.ReadLine()) != null)
{
string[] wordsLine = line.Split(new char[0], StringSplitOptions.RemoveEmptyEntries);
foreach (string str in wordsLine)
{
if (!stopWords.Contains(str.ToLower()))
{
string temp = str.ToLower();
temp = Stemmer.stem(temp);
currentDoc.addTerms(temp);
}
}
}
((MainWindow)System.Windows.Application.Current.MainWindow).ProgressLabel.Content = "Read " + currentDoc.getName();
DocCollection.addDoc(currentDoc);
}
}
catch (Exception ex)
{
System.Windows.Forms.MessageBox.Show("!!!File not read error: " + ex.ToString());
}
}
private void loadDefCollectionButton_Click(object sender, RoutedEventArgs e)
{
DocCollection.cleanCollection();
DocCollection.setupDocCollection();
}