private void btn_predict_Click(object sender, EventArgs e)
{
if (txt_text.Text != "")
{
List <string> CMD = new List <string>();
string s1 = "=== ===Step 0: Văn bản gốc === ===";
CMD.Add(s1);
CMD.Add(" ");
txt_out.Text = "Predicting... ";
int k = 0;
Int32.TryParse(txt_k.Text, out k);
string txt = txt_text.Text;
txt_text.Text = "Predicting... ";
CMD.Add(txt);
string s2 = "=== ===Step 1: Xử lý VnTokenizer === ===";
CMD.Add(" "); CMD.Add(" ");
CMD.Add(s2);
CMD.Add(" ");
File.WriteAllText(".\\temp.txt", txt, Encoding.UTF8);
createBAT("..\\temp.txt", "..\\VnT.txt");
vntokenizer();
string stopword = @".\\data\vnstopword.txt";
string[] StopWord = File.ReadAllLines(stopword);
string output = "";
string[] text = File.ReadAllLines(".\\VnT.txt");
foreach (string textline in text)
{
CMD.Add(textline);
string line = textline.ToLower(); // 1. To Lower
output = removeStopWord(line, output, StopWord);
}
string s3 = "=== ===Step 2: Xóa Stop Word === ===";
CMD.Add(" "); CMD.Add(" ");
CMD.Add(s3);
CMD.Add(" ");
List <string> words = new List <string>();
string word = "";
// find out dictionary and word each file
for (int i = 0; i < output.Length; i++)
{
if (output[i] != ' ')
{
word += output[i];
}
else
{
words.Add(word);
word = "";
}
}
string vb_word = "";
string vb_code = "";
if (rb_knn.Checked)
{
List <int> data = new List <int>();
string[] dictionary = File.ReadAllLines(".\\Dictionary\\Bows.knn.dic");
// create a list same size with dictionary
for (int i = 0; i < dictionary.Length; i++)
{
data.Add(0);
}
// each file, read all word and check in dictionary
foreach (string line in words)
{
vb_word += line + " ";
for (int i = 0; i < dictionary.Length; i++)
{
if (dictionary.ElementAt(i).CompareTo(line) == 0)
{
data[i] += 1;
}
}
}
List <double> Classify = new List <double>();
foreach (int i in data)
{
Classify.Add((double)i * 1.0);
vb_code += i.ToString() + " ";
}
if (k % 2 == 0)
{
k--;
}
Knn kNN_predict = Knn.initialiseKNN(k, ".\\Dictionary\\Dataset.knn.train", true);
string s = kNN_predict.Classify(Classify);
int predict = Int32.Parse(s);
txt_out.Text = "Predict class is: " + get_Name_Class(predict);
}
else
{
List <int> data = new List <int>();
string[] dictionary = File.ReadAllLines(".\\Dictionary\\Bows.svm.dic");
// create a list same size with dictionary
for (int i = 0; i < dictionary.Length; i++)
{
data.Add(0);
}
// each file, read all word and check in dictionary
foreach (string line in words)
{
vb_word += line + " ";
for (int i = 0; i < dictionary.Length; i++)
{
if (dictionary.ElementAt(i).CompareTo(line) == 0)
{
data[i] += 1;
}
}
}
string Classifi = "0";
int index = 0;
foreach (int i in data)
{
Classifi += " " + index.ToString() + ":" + i.ToString();
index++;
vb_code += i.ToString() + " ";
}
File.WriteAllText(".\\temp.txt", Classifi, Encoding.ASCII);
string wri = "cd svm \n svm_predict.exe ..\\temp.txt Model Output.txt";
File.WriteAllText(".\\svm_test.bat", wri, Encoding.ASCII);
Process proc = null;
try
{
string s = Path.GetDirectoryName(Assembly.GetEntryAssembly().Location);
string batDir = string.Format(s);
proc = new Process();
proc.StartInfo.WorkingDirectory = batDir;
proc.StartInfo.FileName = "svm_test.bat";
proc.StartInfo.CreateNoWindow = false;
proc.Start();
proc.WaitForExit();
proc.Close();
}
catch (Exception ex)
{
Console.WriteLine(ex.StackTrace.ToString());
}
string[] result = File.ReadAllLines(".\\svm\\Output.txt");
int predict = Int32.Parse(result[0]);
txt_out.Text = "Predict class is: " + get_Name_Class(predict);
}
CMD.Add(vb_word);
string s4 = "=== ===Step 4: Mã hóa bằng từ điển BOWS === ===";
CMD.Add(" "); CMD.Add(" ");
CMD.Add(s4);
CMD.Add(" ");
CMD.Add(vb_code);
txt_text.Text = String.Join(Environment.NewLine, CMD);
}
else
{
MessageBox.Show("Text box dose not emtry !");
}
}
private void btn_knn_Click(object sender, EventArgs e)
{
Train = false;
txt_text.Text = "Testing is running...";
// Before
if (!File.Exists(".\\Dictionary\\Dataset.knn.test"))
{
SubFolders = Directory.GetDirectories(txt_folderPath.Text);
btn_vntokenizer_Click(sender, e);
Doc2Words();
convertDataset();
}
// Reading model file
int k = 0;
Int32.TryParse(txt_k.Text, out k);
if (k % 2 == 0)
{
k++;
}
Knn kNN_predict = Knn.initialiseKNN(k, ".\\Dictionary\\Dataset.knn.train", true);
// Reading test file
string[] tests = File.ReadAllLines(".\\Dictionary\\Dataset.knn.test");
List <string> Result = new List <string>();
List <string> Label = new List <string>();
List <double> Classify = new List <double>();
foreach (string test in tests)
{
string s = "";
Classify.Clear();
for (int i = 0; i < test.Length; i++)
{
if (test[i] != ',')
{
s += test[i];
}
else
{
double value = 0.0;
Double.TryParse(s, out value);
Classify.Add(value);
s = "";
}
}
Label.Add(s);
string result = kNN_predict.Classify(Classify);
Result.Add(result);
}
List <string> CMD = new List <string>();
string cmd = "";
List <int> T = new List <int>();
List <int> F = new List <int>();
for (int i = 0; i < 16; i++)
{
T.Add(0);
F.Add(0);
}
for (int i = 0; i < Label.Count; i++)
{
if (Label.ElementAt(i) == Result.ElementAt(i))
{
Int32.TryParse(Label.ElementAt(i), out k);
T[k]++;
}
else
{
Int32.TryParse(Label.ElementAt(i), out k);
F[k]++;
}
}
for (int i = 1; i < 16; i++)
{
if (F[i] + T[i] != 0)
{
cmd = "Class: " + get_Name_Class(i) + " :: T = " + T[i].ToString() + " <> F = " + F[i].ToString() + " ";
cmd += "Acc: " + ((double)T[i] * 100 / (F[i] + T[i])).ToString() + " %";
CMD.Add(cmd);
}
}
cmd = "Test Accuracy: " + ((double)T.Sum() * 100 / (T.Sum() + F.Sum())).ToString() + " %";
CMD.Add(cmd);
txt_text.Text = String.Join(Environment.NewLine, CMD);
// After
Train = true;
MessageBox.Show("Done !");
}
