public static List <string> DoDetect(string input)
{
input = NGramKeyword.removeHTMLWithLineAndSpace(input);
List <string> R = new List <string>();
string[] _sp = stringSeparate(input);
for (int k = 0; k < _sp.Length; k++)
{
string[] sp = NormalizeStringToExtract(_sp[k]).Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
string cache = "";
int c = 0;
for (int i = 0; i < sp.Length; i++)
{
string cacheTrim = cache.Trim(new char[] { ' ', '.', ',', '…' });
if (IsCapital(sp[i]))
{
cache += sp[i] + " ";
c++;
if (IsCode(sp[i]))
{
R.Add(sp[i].Trim(new char[] { ' ', '.', ',', '…' }));
}
}
else
{
if (c > 2)
{
R.Add(cacheTrim);
}
else if (ForeignDetect(cache))
{
R.Add(cacheTrim);
}
cache = "";
c = 0;
}
if (i == sp.Length - 1)
{
if (c > 2)
{
R.Add(cacheTrim);
}
else if (ForeignDetect(cache))
{
R.Add(cacheTrim);
}
cache = "";
c = 0;
}
}
}
var rx = R.Distinct().ToList();
rx.Remove("");
rx.Remove("AloBacsi");
return(rx);
}
public static List <string> ExtractKeywordBasic(string input)
{
input = NGramKeyword.removeHTMLWithLineAndSpace(input);
List <string> R = new List <string>();
string[] _sp = stringSeparate(input);
for (int k = 0; k < _sp.Length; k++)
{
string[] sp = NormalizeStringToExtract(_sp[k]).Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
R.AddRange(sp);
}
var rx = R.Distinct().ToList();
return(rx);
}
