public static void tienXuLy(string input, string output)
{
EnglishPorter2Stemmer stemword = new EnglishPorter2Stemmer();
string line = null;
List <string> Texts = new List <string>();
//Doc cac van ban tu file input
using (StreamReader filein = new StreamReader(input))
{
while ((line = filein.ReadLine()) != null)
{
Texts.Add(line);
}
filein.Close();
}
//Console.WriteLine("===============Cac dong trong file input===============");
//foreach (string t in Texts)
//{
// Console.WriteLine(t);
//}
//Console.WriteLine("===============Cac dong trong file output===============");
using (StreamWriter fileout = new StreamWriter(output))
{
for (int i = 0; i < Texts.Count(); i++)
{
Texts[i] = Texts[i].ToLower();
Texts[i] = XoaKyTu(Texts[i]);
Texts[i] = XoaTrang(Texts[i]);
Texts[i] = BoStopWords(Texts[i]);
Texts[i] = PorterStemming(Texts[i]);
if (i == (Texts.Count - 1))
{
// Console.WriteLine(Texts[i]);
fileout.Write(Texts[i]);
}
else
{
//Console.WriteLine(Texts[i]);
fileout.WriteLine(Texts[i]);
}
}
fileout.Close();
}
}
public static string PorterStemming(string str)
{
EnglishPorter2Stemmer stemword = new EnglishPorter2Stemmer();
string templine = null;
List <string> StopWords = new List <string>();
string[] temp = str.Split(' ');
for (int i = 0; i < temp.Length; i++)
{
temp[i] = stemword.Stem(temp[i]).Value;
if (i == (temp.Length - 1))
{
templine = templine + temp[i];
}
else
{
templine = templine + temp[i] + " ";
}
}
return(templine);
}
public static string PorterStemming(string str)
{
EnglishPorter2Stemmer stemword = new EnglishPorter2Stemmer();
StringBuilder templine = new StringBuilder();
List <string> StopWords = new List <string>();
string[] temp = str.Split(' ');
for (int i = 0; i < temp.Length; i++)
{
temp[i] = stemword.Stem(temp[i]).Value;
if (i == (temp.Length - 1))
{
templine.Append(temp[i]);
}
else
{
templine.Append(temp[i] + " ");
}
}
return(templine.ToString());
}