//Get StopWord
private void button8_Click(object sender, EventArgs e)
{
var dr = openFileDialog2.ShowDialog();
if (dr != DialogResult.OK)
{
return;
}
foreach (var unused in openFileDialog1.FileNames)
{
string content;
using (var reader = new StreamReader(openFileDialog2.FileName, Encoding.UTF8))
{
content = reader.ReadToEnd();
}
Stopword = new Stopword(content);
if (content != "")
{
lbTxt.Text = @"đã có stopword";
_sword = true;
}
else
{
MessageBox.Show(@"File rổng chọn lại", @"Select File Stopword", MessageBoxButtons.OK,
MessageBoxIcon.Stop);
_sword = false;
lbTxt.Text = @"chưa có stopword";
}
}
}
public document(string title, string content, Stopword st, string rg, int s)
{
Title = title;
Content = content.ToLower();
IStemmer stemmer = new EnglishStemmer();
var regex = "";
if (rg == null)
{
regex = @"[A-Za-z\-]+";
}
else
{
regex = rg;
}
var valueEnumerable = Regex.Matches(content, regex);
var lt = valueEnumerable.Cast <Match>().Select(match => match.Value).ToList();
count = lt.Count;
StopWords = st.Lst.ToArray();
ListWorld = valueEnumerable.Cast <Match>().Select(match => match.Value).
ToList().ConvertAll(a => a.ToLower()).Except(StopWords).OrderBy(a => a);
ListWorld = ListWorld.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
public Form1()
{
InitializeComponent();
DTable = new DataTable();
openFileDialog1.Multiselect = true;
openFileDialog5.Multiselect = false;
Stopword = null;
}
public Query(string s, Stopword sw, int e, string rg)
{
Lstq = new List <Tuple <string, double> >();
var b = s.Split(new string[] { "\t", "\t" }, StringSplitOptions.None);
if (b.Length != 2)
{
return;
}
IdQuery = b[0];
Title = b[1];
var regex = "";
if (rg == null)
{
regex = @"[A-Za-z\-]+";
}
else
{
regex = rg;
}
IStemmer stemmer = new EnglishStemmer();
var valueEnumerable = Regex.Matches(Title = b[1], regex);
IEnumerable <string> Lst = null;
try
{
Lst = valueEnumerable.Cast <Match>().Select(match => match.Value).ToList().Except(sw.Lst.ToArray()).OrderBy(a => a);
Lst = Lst.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
catch (Exception)
{
Lst = valueEnumerable.Cast <Match>().Select(match => match.Value).ToList().OrderBy(a => a);
Lst = Lst.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
Dictionary <string, int> counts = Lst.GroupBy(x => x)
.ToDictionary(g => g.Key,
g => g.Count());
var ls = Lst.Distinct().ToList();
foreach (var item in counts)
{
Lstq.Add(new Tuple <string, double>(item.Key, (double)item.Value / (double)ls.Count));
}
}
public QueryAON(string s, Stopword sw)
{
lst = new List <Tuple <string, int> >();
const string regex = @"[A-Za-z\-()]+";
IStemmer stemmer = new EnglishStemmer();
var valueEnumerable = Regex.Matches(s, regex);
Lst = new List <string>();
Lst1 = new List <string>();
var Lstq = new List <Tuple <string, double> >();
if (sw != null)
{
var dt = sw.Lst.Where(a => a != "AND" && a != "OR" && a != "NOT").ToList();
Lst = valueEnumerable.Cast <Match>().Select(match => match.Value).ToList().Except(dt.ToArray()).ToList();
Lst = Lst.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
else
{
Lst = valueEnumerable.Cast <Match>().Select(match => match.Value).ToList();
Lst = Lst.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
var i = 0;
for (var j = 0; j < Lst.Count; j++)
{
{
if (Lst[j] == "and" || Lst[j] == "or" || Lst[j] == "not" || Lst[j] == "(" || Lst[j] == ")")
{
continue;
}
lst.Add(new Tuple <string, int>(Lst[j], i));
Lst1.Add(Lst[j]);
Lst[j] = i.ToString();
i++;
}
}
}
public Query(string s, Stopword sw, string rg)
{
var regex = "";
if (rg == null)
{
regex = @"[A-Za-z\-]+";
}
else
{
regex = rg;
}
IStemmer stemmer = new EnglishStemmer();
s = s.ToLower();
var valueEnumerable = Regex.Matches(s, regex);
IEnumerable <string> Lst = null;
Lstq = new List <Tuple <string, double> >();
if (sw != null)
{
Lst = valueEnumerable.Cast <Match>().Select(match => match.Value).ToList().Except(sw.Lst.ToArray()).OrderBy(a => a);
Lst = Lst.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
else
{
Lst = valueEnumerable.Cast <Match>().Select(match => match.Value).OrderBy(a => a).ToList();
Lst = Lst.ToList().ConvertAll(d => stemmer.Stem(d.ToLower()));
}
var ls = Lst.Distinct().ToList();
foreach (var item in ls)
{
var l = Lst.Count(a => a == item);
Lstq.Add(new Tuple <string, double>(item, (double)l / Lst.ToList().Count));
}
}
