private void StandardizationButton_Click(object sender, EventArgs e)
{
ClossButton();
MyData.TheDoc = MyFun.DocStandardization(MyData.TheDoc);
StatusLabel.Text = "Document standardization process has been completed.";
OpenButton();
}
private void RemoveStopButton_Click(object sender, EventArgs e)
{
ClossButton();
MessageBox.Show("Tip: If the document is longer, it may take you few minutes, please be patient ...", "Prompt", 0, MessageBoxIcon.Asterisk);
MyData.TheDoc = MyFun.RemoveStop(MyData.TheDoc);
StatusLabel.Text = "Remove stop-words has been completed.";
OpenButton();
}
private void DoWork()
{
lock (this)
{
ClossButton();
DateTime dt1 = DateTime.Now;
StatusLabel.Text = "Keyword extraction process is ongoing:0%";
ResultListView.Items.Clear();
MyData.WordsFre = MyFun.StatisticsWords(MyData.TheDoc);
for (int i = 0; i < MyData.WordsFre.Length; i++)
{
MyData.WordsFre[i].EntropyDifference_Max();
progressBar1.Value = i * 100 / MyData.WordsFre.Length;
StatusLabel.Text = "Keyword extraction process is ongoing: " + progressBar1.Value + "%";
}
MyFun.QuickSort(MyData.WordsFre, 0, MyData.WordsFre.Length - 1);
int WordsNum = 0;
for (int i = 0; i < MyData.WordsFre.Length; i++)
{
if (MyData.WordsFre[i].ED > 0)
{
WordsNum++;
}
else
{
break;
}
}
ListViewItem[] lvi = new ListViewItem[WordsNum];
for (int i = 0; i < WordsNum; i++)
{
lvi[i] = new ListViewItem();
lvi[i].SubItems[0].Text = (i + 1).ToString();
lvi[i].SubItems.Add(MyData.WordsFre[i].Word.ToString());
lvi[i].SubItems.Add(MyData.WordsFre[i].ED.ToString());
lvi[i].SubItems.Add(MyData.WordsFre[i].Frequency.ToString());
}
ResultListView.Items.AddRange(lvi);
DateTime dt2 = DateTime.Now;
SaveButton.Enabled = true;
progressBar1.Value = 100;
StatusLabel.Text = "Keyword extraction has been completed. The extraction spend " + (dt2 - dt1).ToString() + ".";
OpenButton();
}
}
private void ParticipleButton_Click(object sender, EventArgs e)
{
ClossButton();
MessageBox.Show("提示：如果文档较长，可能需要一些时间，请耐心等待...", "提示", 0, MessageBoxIcon.Asterisk);
// 系统初始化
if (!MyFun.ICTCLAS_Init(null))
{
System.Console.WriteLine("Init ICTCLAS failed!");
System.Console.Read();
}
//导入用户字典
string path = @"userdic.txt";
Encoding encode = Encoding.GetEncoding("GB2312");
string userdic = File.ReadAllText(path, encode);
MyFun.ICTCLAS_ImportUserDict(userdic, userdic.Length, eCodeType.CODE_TYPE_UNKNOWN);
int count = MyData.TheDoc.Length; //先得到结果的词数
result_t[] result = new result_t[count]; //在客户端申请资源
int i = 0;
//对字符串进行分词处理
int nWrdCnt = MyFun.ICTCLAS_ParagraphProcessAW(MyData.TheDoc, result, eCodeType.CODE_TYPE_GB, 1);
result_t r;
//取字符串真实长度
byte[] mybyte = System.Text.Encoding.Default.GetBytes(MyData.TheDoc);
byte[] byteWord;
string TheDoc = "";
for (i = 0; i < nWrdCnt; i++)
{
r = result[i];
byteWord = new byte[r.length];
//取字符串一部分
Array.Copy(mybyte, r.start, byteWord, 0, r.length);
string wrd = System.Text.Encoding.Default.GetString(byteWord);
if (wrd == " ")
{
continue;
}
if (i == 0)
{
TheDoc += wrd;
}
else
{
TheDoc += " " + wrd;
}
}
//释放资源退出
MyFun.ICTCLAS_Exit();
MyData.TheDoc = TheDoc;
StatusLabel.Text = "中文分词已经完成。";
OpenButton();
}
