private void buttonClasterize_Click(object sender, EventArgs e)
{
var text = _task.Description;
var morphLib = new TA.Morph.MorphLib(text);
var firstFormWords = morphLib.ToMainForm("S", "A");
var firstFormString = string.Join(" ", firstFormWords);
var method1 = new TA.Statistic.MutualInformation(firstFormWords);
var method2 = new TA.Statistic.TScore(firstFormWords);
var method3 = new TA.Statistic.LogLikelihood(firstFormWords);
var clasterItems = new List<Claster.ClasterItem>();
for (int i = 0; i < firstFormWords.Count - 1; i++)
{
var clasterItemText = firstFormWords[i] + " " + firstFormWords[i + 1];
if (clasterItems.Any(j => j.Item.ToString() == clasterItemText))
continue;
var m1 = method1.Calculate(clasterItemText);
var m2 = method2.Calculate(clasterItemText);
var m3 = method3.Calculate(clasterItemText);
var clasterItem = new Claster.ClasterItem(3) { Item = clasterItemText };
clasterItem.SetValue(0, m1);
clasterItem.SetValue(1, m2);
clasterItem.SetValue(2, m3);
clasterItems.Add(clasterItem);
}
var clasterizeBigrams = new TA.Claster.ClasterizeBigram(text, clasterItems);
clasterizeBigrams.Run();
var topTerms = clasterizeBigrams.GetTerms(50);
var topNotTerms = clasterizeBigrams.GetNotTerms(50);
Logger.LogText(string.Format("\r\nТОП {0} терминов:", topTerms.Count));
Logger.LogHtml(string.Format("<h2>ТОП {0} терминов:</h2><ul>", topTerms.Count));
foreach(var term in topTerms)
{
Logger.LogText(term);
Logger.LogHtml(string.Format("<li>{0}</li>", term));
}
Logger.LogHtml("</ul>");
Logger.LogText(string.Format("\r\nТОП {0} не терминов:", topNotTerms.Count));
Logger.LogHtml(string.Format("<h2>ТОП {0} не терминов:</h2><ul>", topNotTerms.Count));
foreach (var notTerm in topNotTerms)
{
Logger.LogText(notTerm);
Logger.LogHtml(string.Format("<li>{0}</li>", notTerm));
}
resultTextBox.Text += "\r\n";
Logger.LogHtml("</ul>");
}
private void buttonTScore_Click(object sender, EventArgs e)
{
// check format
if (!isTwoWords())
{
MessageBox.Show("Неверный формат входной строки.", this.Text, MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
// get first form of source string
string sourceString = sourceFirstForm();
// get result
var tScore = new TA.Statistic.TScore(_words);
var result = tScore.Calculate(sourceString);
// print result
Logger.LogText(string.Format("Метод T-Score [{0}]: {1}.", sourceString, result));
Logger.LogHtml(string.Format("<p><strong>Метод T-Score [{0}]:</strong> {1}.</p>", sourceString, result));
}
private void buttonClasterize_Click(object sender, EventArgs e)
{
var text = _task.Description;
var morphLib = new TA.Morph.MorphLib(text);
var firstFormWords = morphLib.ToMainForm("S", "A");
var firstFormString = string.Join(" ", firstFormWords);
var method1 = new TA.Statistic.MutualInformation(firstFormWords);
var method2 = new TA.Statistic.TScore(firstFormWords);
var method3 = new TA.Statistic.LogLikelihood(firstFormWords);
var clasterItems = new List <Claster.ClasterItem>();
for (int i = 0; i < firstFormWords.Count - 1; i++)
{
var clasterItemText = firstFormWords[i] + " " + firstFormWords[i + 1];
if (clasterItems.Any(j => j.Item.ToString() == clasterItemText))
{
continue;
}
var m1 = method1.Calculate(clasterItemText);
var m2 = method2.Calculate(clasterItemText);
var m3 = method3.Calculate(clasterItemText);
var clasterItem = new Claster.ClasterItem(3)
{
Item = clasterItemText
};
clasterItem.SetValue(0, m1);
clasterItem.SetValue(1, m2);
clasterItem.SetValue(2, m3);
clasterItems.Add(clasterItem);
}
var clasterizeBigrams = new TA.Claster.ClasterizeBigram(text, clasterItems);
clasterizeBigrams.Run();
var topTerms = clasterizeBigrams.GetTerms(50);
var topNotTerms = clasterizeBigrams.GetNotTerms(50);
Logger.LogText(string.Format("\r\nТОП {0} терминов:", topTerms.Count));
Logger.LogHtml(string.Format("<h2>ТОП {0} терминов:</h2><ul>", topTerms.Count));
foreach (var term in topTerms)
{
Logger.LogText(term);
Logger.LogHtml(string.Format("<li>{0}</li>", term));
}
Logger.LogHtml("</ul>");
Logger.LogText(string.Format("\r\nТОП {0} не терминов:", topNotTerms.Count));
Logger.LogHtml(string.Format("<h2>ТОП {0} не терминов:</h2><ul>", topNotTerms.Count));
foreach (var notTerm in topNotTerms)
{
Logger.LogText(notTerm);
Logger.LogHtml(string.Format("<li>{0}</li>", notTerm));
}
resultTextBox.Text += "\r\n";
Logger.LogHtml("</ul>");
}
private void buttonTScore_Click(object sender, EventArgs e)
{
// check format
if (!isTwoWords())
{
MessageBox.Show("Неверный формат входной строки.", this.Text, MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
// get first form of source string
string sourceString = sourceFirstForm();
// get result
var tScore = new TA.Statistic.TScore(_words);
var result = tScore.Calculate(sourceString);
// print result
Logger.LogText(string.Format("Метод T-Score [{0}]: {1}.", sourceString, result));
Logger.LogHtml(string.Format("<p><strong>Метод T-Score [{0}]:</strong> {1}.</p>", sourceString, result));
}