public int LoadBaiViet()
{
List<BAIVIET_getall_newResult> lt = new List<BAIVIET_getall_newResult>();
BUSBaiViet BUSBaiViet = new BUSBaiViet();
lt = BUSBaiViet.SelectBAIVIETsAll_new();
LOAIBAIVIET ltLoaiBaiViet = new LOAIBAIVIET();
BUSLoaiBaiViet BUSLoaiBaiViet = new BUSLoaiBaiViet();
if (lt.Count > 0)
{
lt.Sort(SortByDate);
this.GridViewBaiViet.DataSource = lt;
GridViewBaiViet.DataBind();
PanelDanhSach.Visible = true;
PanelMessage.Visible = false;
// for (int i = 0; i < lt.Count; i++)
// {
// GridViewBaiViet.Rows[i].Cells[4].Text = BUSLoaiBaiViet.TimKiem((int)lt[i].MaLoaiBaiViet).TenLoaiBaiViet;
// }
return lt.Count;
}
else
{
PanelDanhSach.Visible = false;
PanelMessage.Visible = true;
return 0;
}
}
private void bt_clustering_Click(object sender, EventArgs e)
{
List<string> initialContent = ReadData("C:\\Users\\LENOVO\\Desktop\\回复提取3.0.txt");
List<string> content = new List<string>();
for (int i = 0; i < initialContent.Count; i++)
{
List<string> matchs=GetMatchs(@"(?<=\b.+?\b\s\b\w{4}\d\b\s).*", initialContent[i]);
if (matchs.Count != 0)
content.Add(matchs[0]);
}
List<string> segmentWords = GetSegmentWords(content);
string[] stopWords = File.ReadAllLines(@"中文停用词表（专用）.txt", System.Text.Encoding.Default);
List<theWord> theWords = new List<theWord>();
foreach (string s in segmentWords.Distinct().ToList())
{
int i = 0;
foreach (string word in stopWords)
{
if (word == s)
{
i = 1;
break;
}
}
if (!Regex.IsMatch(s, @"[\u4e00-\u9fbb]+$"))//去除非汉字
i = 1;
if (i == 1)
continue;
int times = 0;
foreach (string word in segmentWords)
{
if (s == word)
times++;
}
theWord theword;
theword.text = s;
theword.times = times;
theWords.Add(theword);
}
theWords.Sort((left, right) =>
{
if (left.times > right.times)
return -1;
else if (left.times == right.times)
return 0;
else
return 1;
});
for (int i = 0; i < theWords.Count; i++)
{
tb_content.Text += theWords[i].text + "(" + theWords[i].times + ")" + "\r\n";
}
StreamWriter sw = new StreamWriter("C:\\Users\\LENOVO\\Desktop\\流行词排行3.0.txt");
sw.Write(tb_content.Text);
sw.Close();
}