/// <summary>
/// Detects the language by finding the top french words, and top english
/// PRECONDITION: french stem and english stem must have been run at least once
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void btnDetectLanguage_Click(object sender, EventArgs e)
{
//get the most popular french words
IEnumerable<string> topFrench = GetMostPopularWords(_frenchStemmedCounts);
//get the most popular english words
IEnumerable<string> topEnglish = GetMostPopularWords(_englishStemmedCounts);
int frenchCounter = 0;
int englishCounter = 0;
FrenchStemmer frenchStemmer = new FrenchStemmer();
PorterStemmer englishStemmer = new PorterStemmer();
//find the top french in text
string input = this.txtInput.Text;
foreach (string term in input.Split(' '))
{
string frenchStem = frenchStemmer.stemTerm(term);
string englishStem = englishStemmer.stemTerm(term);
if (string.IsNullOrEmpty(term))
continue;
foreach (string frenchTerm in topFrench)
{
// if (frenchTerm.IgnoringCaseEquals(term))
if (frenchTerm.IgnoringCaseEquals(frenchStem))
frenchCounter++;
}
foreach (string englishTerm in topEnglish)
{
// if (englishTerm.IgnoringCaseEquals(term))
if (englishTerm.IgnoringCaseEquals(englishStem))
englishCounter++;
}
}
if (frenchCounter > englishCounter)
MessageBox.Show("French");
else if (englishCounter > frenchCounter)
MessageBox.Show("English");
else
MessageBox.Show("Not sure, have you run a stem on each language to train it?");
}
/// <summary>
/// Does a stemming of the text in english, with step by step output.
/// </summary>
private void btnEnglishStem_Click(object sender, EventArgs e)
{
string input = txtInput.Text;
this.txtOutput.Text = "";
this.txtFinalText.Text = "";
foreach (string term in input.Split(' '))
{
if (string.IsNullOrEmpty(term))
continue;
//check that its actually goign to do anything
PorterStemmer preStemmer = new PorterStemmer();
string stemmedTerm = preStemmer.stemTerm(term);
//save the stemmed term to the stemmed counts
// do this before you skip it if its already done
if (_englishStemmedCounts.ContainsKey(stemmedTerm))
{
_englishStemmedCounts[stemmedTerm]++;
}
else
{
_englishStemmedCounts.Add(stemmedTerm, 1);
}
txtFinalText.Text += stemmedTerm + " ";
if (term.Equals(stemmedTerm))
continue;
//expand out the steps in full
PorterStemmer stemmer = new PorterStemmer();
stemmer.setTerm(term);
stemmer.PreStem();
// txtOutput.Text += term + @" -> " + stemmer.stemTerm(term) + Environment.NewLine;
stemmer.Step1();
txtOutput.Text += term + " Step 1 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step2();
txtOutput.Text += term + " Step 2 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step3();
txtOutput.Text += term + " Step 3 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step4();
txtOutput.Text += term + " Step 4 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step5();
txtOutput.Text += term + " Step 5 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step6();
txtOutput.Text += term + " Step 6 " + stemmer.getCurentstate() + Environment.NewLine;
txtOutput.Text += Environment.NewLine;
stemmer.PostStem();
txtOutput.Text += "Final Stem: " + stemmer.getTerm() + Environment.NewLine + Environment.NewLine;
}
}
/// <summary>
/// Does a stemming of the text in english, with step by step output.
/// </summary>
private void btnEnglishStem_Click(object sender, EventArgs e)
{
string input = txtInput.Text;
this.txtOutput.Text = "";
this.txtFinalText.Text = "";
foreach (string term in input.Split(' '))
{
if (string.IsNullOrEmpty(term))
{
continue;
}
//check that its actually goign to do anything
PorterStemmer preStemmer = new PorterStemmer();
string stemmedTerm = preStemmer.stemTerm(term);
//save the stemmed term to the stemmed counts
// do this before you skip it if its already done
if (_englishStemmedCounts.ContainsKey(stemmedTerm))
{
_englishStemmedCounts[stemmedTerm]++;
}
else
{
_englishStemmedCounts.Add(stemmedTerm, 1);
}
txtFinalText.Text += stemmedTerm + " ";
if (term.Equals(stemmedTerm))
{
continue;
}
//expand out the steps in full
PorterStemmer stemmer = new PorterStemmer();
stemmer.setTerm(term);
stemmer.PreStem();
// txtOutput.Text += term + @" -> " + stemmer.stemTerm(term) + Environment.NewLine;
stemmer.Step1();
txtOutput.Text += term + " Step 1 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step2();
txtOutput.Text += term + " Step 2 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step3();
txtOutput.Text += term + " Step 3 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step4();
txtOutput.Text += term + " Step 4 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step5();
txtOutput.Text += term + " Step 5 " + stemmer.getCurentstate() + Environment.NewLine;
stemmer.Step6();
txtOutput.Text += term + " Step 6 " + stemmer.getCurentstate() + Environment.NewLine;
txtOutput.Text += Environment.NewLine;
stemmer.PostStem();
txtOutput.Text += "Final Stem: " + stemmer.getTerm() + Environment.NewLine + Environment.NewLine;
}
}