public void readdata()
{
Segment segment = new Segment();
ICollection <WordInfo> words;
ICollection <WordInfo> re_words;
SinaJSON item;
SinaJSON re_item;
List <string> ls = new List <string>();
string[] sp = Datas.ads_word.Split(',');
int quan;
int t;
int i;
double temp_quan;
List <ju_word> te;//test
Datas.auto_read.WaitOne();
for (; Datas.now_readnum < Datas.temp_data.Count; Datas.now_readnum++)
{
if (Datas.now_readnum == Datas.temp_data.Count)
{
Datas.auto_read.WaitOne();
}
quan = 0;
item = Datas.temp_data.ElementAt(Datas.now_readnum);
for (i = 0; i < sp.Length; i++)
{
t = item.Text.IndexOf(sp[i]);
if (t != -1)
{
quan++;
}
}
if (quan < 4)
{
item.Text = Regex.Replace(item.Text, @"[a-zA-Z]+[:/]*[a-zA-Z\d\./]+", "");
for (i = item.Text.Length - 1; i > -1; i--)
{
if (item.Text[i] == 55356 || item.Text[i] == 55357)
{
item.Text = item.Text.Remove(i);
}
}
words = segment.DoSegment(item.Text);
ls.Clear();
if (words.Count != 0)
{
foreach (WordInfo wordInfo in words)
{
if (!ls.Contains(wordInfo.Word))
{
ls.Add(wordInfo.Word);
if (Datas.frequency[6].ContainsKey(wordInfo.Word))
{
Datas.frequency[6][wordInfo.Word].now_nu++;
Datas.frequency[6][wordInfo.Word].position.Add(Datas.now_time_count);
}
else
{
words temp = new words(wordInfo.Word, Datas.now_time_count);
Datas.frequency[6].Add(wordInfo.Word, temp);
}
}
}
}
Datas.total_data[6].Add(item);
Datas.total_num++;
Datas.now_time_count++;
if (Datas.now_time_count >= Datas.now_time_max)
{
Datas.frequency[Datas.now_time_read % 6] = Datas.frequency[6];
Datas.total_data[Datas.now_time_read % 6] = Datas.total_data[6];
Datas.frequency[6] = new Dictionary <string, 微博舆论.words>();
Datas.total_data[6] = new List <SinaJSON>();
//if(Datas.now_time_read>0)
//算基础权重
for (i = 0; i < Datas.total_data[Datas.now_time_read % 6].Count; i++)//~~
{
string max;
re_item = Datas.total_data[Datas.now_time_read % 6].ElementAt(i);//~~
re_words = segment.DoSegment(re_item.Text);
//max = max_ocur(re_words,Datas.frequency[Datas.now_time_read % 6],i);
if (re_words.Count != 0)
{
max = re_words.ElementAt(0).Word;
foreach (WordInfo word in re_words)
{
if (Datas.frequency[Datas.now_time_read % 6].ContainsKey(word.Word))
{
if (Datas.frequency[Datas.now_time_read % 6][max].now_nu < Datas.frequency[Datas.now_time_read % 6][word.Word].now_nu)
{
max = word.Word;
}
}
}
foreach (WordInfo temp in re_words)
{
temp_quan = (0.5 + 0.5 * Datas.frequency[Datas.now_time_read % 6][temp.Word].now_nu / Datas.frequency[Datas.now_time_read % 6][max].now_nu);
Datas.frequency[Datas.now_time_read % 6][temp.Word].weight_quan += (temp_quan * (Datas.weight_word * re_item.Attitudes_count + (1 - Datas.weight_word) * re_item.Comments_count));
}
}
}
//算bursty 待加用户权重
List <string> l_s = new List <string>();
List <int> l_i = new List <int>();
List <ju_word> l_julei;
int j;
if (Datas.frequency[(Datas.now_time_read + 5) % 6].Count > 0)
{
l_s.Clear(); //突发度大于0
l_i.Clear(); //在字典中的位置
for (i = 0; i < Datas.frequency[Datas.now_time_read % 6].Count; i++)
{
double wehit_cou = 0;
if (Datas.frequency[(Datas.now_time_read + 5) % 6].ContainsKey(Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Key))
{
wehit_cou = Datas.frequency[(Datas.now_time_read + 5) % 6][Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Key].weight_quan;
}
Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Value.bursty =
(Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Value.weight_quan - wehit_cou) / 2;
if (Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Value.bursty > 0)
{
l_s.Add(Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Key);
l_i.Add(i);
}
}
//距离聚类
l_julei = new List <ju_word>();
for (i = 0; i < l_s.Count; i++)
{
ju_word ju_t = new ju_word(l_s.ElementAt(i), Datas.frequency[Datas.now_time_read % 6][l_s.ElementAt(i)].position);
l_julei.Add(ju_t);
}
double f_z = 0;
double ju_temp;
double ju_min;
double fengzi = 0;
double fenmu = 0;
int temp_count;//the nunmber of the same
int ju_x = 0;
int ju_y = 0;
List <int> l_posi = new List <int>();
while (f_z < 15)
{
ju_min = 10000;
ju_x = 0;
ju_y = 0;
List <int> l_temp_pos = new List <int>();
for (i = 0; i < l_julei.Count; i++)
{
for (j = i + 1; j < l_julei.Count; j++)
{
//zuo = l_julei.ElementAt(i).Split(',');
//you = l_julei.ElementAt(j).Split(',');
for (int x = 0; x < l_julei.ElementAt(i).word.Count; x++)
{
for (int y = 0; y < l_julei.ElementAt(j).word.Count; y++)
{
fengzi += (Datas.frequency[Datas.now_time_read % 6][l_julei.ElementAt(i).word.ElementAt(x)].bursty
* Datas.frequency[Datas.now_time_read % 6][l_julei.ElementAt(j).word.ElementAt(y)].bursty);
}
}
fengzi = fengzi / (l_julei.ElementAt(i).word.Count + l_julei.ElementAt(i).word.Count);
temp_count = 0;
for (int ju_i = 0; ju_i < l_julei.ElementAt(i).position.Count; ju_i++)
{
if (l_julei.ElementAt(j).position.Contains(l_julei.ElementAt(i).position.ElementAt(ju_i)))
{
temp_count++;
l_temp_pos.Add(l_julei.ElementAt(i).position.ElementAt(ju_i));
}
}
fenmu = (double)temp_count / (double)Datas.now_time_max;
ju_temp = fengzi / fenmu;
if (ju_temp < ju_min)
{
ju_min = ju_temp;
ju_x = i;
ju_y = j;
l_posi = new List <int>(l_temp_pos);
}
fengzi = 0;
fenmu = 0;
l_temp_pos.Clear();
}
te = l_julei.OrderBy(ju_word => ju_word.word.Count).ToList();
}
if (ju_x == 0 && ju_y == 0)
{
break;
}
//聚类一个
for (i = 0; i < l_julei.ElementAt(ju_y).word.Count; i++)
{
l_julei.ElementAt(ju_x).word.Add(l_julei.ElementAt(ju_y).word.ElementAt(i));
}
l_julei.ElementAt(ju_x).position = new List <int>(l_posi);
l_julei.RemoveAt(ju_y);
f_z = ju_min;
}
List <ju_word> temp = new List <ju_word>();
for (i = 0; i < l_julei.Count; i++)
{
if (l_julei.ElementAt(i).word.Count > 1)
{
temp.Add(l_julei.ElementAt(i));
}
}
Datas.ju_list[Datas.now_time_read % 6] = temp.OrderByDescending(ju_word => ju_word.position.Count).ToList();
//this.Invoke(new updateListboxFreq(doUpdateListboxFreq), new object[] { Datas.ju_list });
if (!Datas.ProgressBarCompelet)
{
Datas.ProgressBarCompelet = true;
}
}
Datas.now_time_count = 0;
Datas.now_time_read++;
}
}
}
}
public void readdata()
{
Segment segment = new Segment();
ICollection<WordInfo> words;
ICollection<WordInfo> re_words;
SinaJSON item;
SinaJSON re_item;
List<string> ls = new List<string>();
string[] sp = Datas.ads_word.Split(',');
int quan;
int t;
int i;
double temp_quan;
List<ju_word> te;//test
Datas.auto_read.WaitOne();
for (; Datas.now_readnum < Datas.temp_data.Count; Datas.now_readnum++)
{
if (Datas.now_readnum == Datas.temp_data.Count)
Datas.auto_read.WaitOne();
quan = 0;
item = Datas.temp_data.ElementAt(Datas.now_readnum);
for (i = 0; i < sp.Length; i++)
{
t = item.Text.IndexOf(sp[i]);
if (t != -1)
quan++;
}
if (quan < 4)
{
item.Text = Regex.Replace(item.Text, @"[a-zA-Z]+[:/]*[a-zA-Z\d\./]+", "");
for (i = item.Text.Length - 1; i > -1; i--)
{
if (item.Text[i] == 55356 || item.Text[i] == 55357)
{
item.Text = item.Text.Remove(i);
}
}
words = segment.DoSegment(item.Text);
ls.Clear();
if (words.Count != 0)
{
foreach (WordInfo wordInfo in words)
{
if (!ls.Contains(wordInfo.Word))
{
ls.Add(wordInfo.Word);
if (Datas.frequency[6].ContainsKey(wordInfo.Word))
{
Datas.frequency[6][wordInfo.Word].now_nu++;
Datas.frequency[6][wordInfo.Word].position.Add(Datas.now_time_count);
}
else
{
words temp = new words(wordInfo.Word, Datas.now_time_count);
Datas.frequency[6].Add(wordInfo.Word, temp);
}
}
}
}
Datas.total_data[6].Add(item);
Datas.total_num++;
Datas.now_time_count++;
if (Datas.now_time_count >= Datas.now_time_max)
{
Datas.frequency[Datas.now_time_read % 6] = Datas.frequency[6];
Datas.total_data[Datas.now_time_read % 6] = Datas.total_data[6];
Datas.frequency[6] = new Dictionary<string, 微博舆论.words>();
Datas.total_data[6] = new List<SinaJSON>();
//if(Datas.now_time_read>0)
//算基础权重
for (i = 0; i < Datas.total_data[Datas.now_time_read % 6].Count; i++)//~~
{
string max;
re_item = Datas.total_data[Datas.now_time_read % 6].ElementAt(i);//~~
re_words = segment.DoSegment(re_item.Text);
//max = max_ocur(re_words,Datas.frequency[Datas.now_time_read % 6],i);
if (re_words.Count != 0)
{
max = re_words.ElementAt(0).Word;
foreach (WordInfo word in re_words)
{
if (Datas.frequency[Datas.now_time_read % 6].ContainsKey(word.Word))
{
if (Datas.frequency[Datas.now_time_read % 6][max].now_nu < Datas.frequency[Datas.now_time_read % 6][word.Word].now_nu)
max = word.Word;
}
}
foreach (WordInfo temp in re_words)
{
temp_quan = (0.5 + 0.5 * Datas.frequency[Datas.now_time_read % 6][temp.Word].now_nu / Datas.frequency[Datas.now_time_read % 6][max].now_nu);
Datas.frequency[Datas.now_time_read % 6][temp.Word].weight_quan += (temp_quan * (Datas.weight_word * re_item.Attitudes_count + (1 - Datas.weight_word) * re_item.Comments_count));
}
}
}
//算bursty 待加用户权重
List<string> l_s = new List<string>();
List<int> l_i = new List<int>();
List<ju_word> l_julei;
int j;
if (Datas.frequency[(Datas.now_time_read + 5) % 6].Count > 0)
{
l_s.Clear();//突发度大于0
l_i.Clear();//在字典中的位置
for (i = 0; i < Datas.frequency[Datas.now_time_read % 6].Count; i++)
{
double wehit_cou = 0;
if (Datas.frequency[(Datas.now_time_read + 5) % 6].ContainsKey(Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Key))
wehit_cou = Datas.frequency[(Datas.now_time_read + 5) % 6][Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Key].weight_quan;
Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Value.bursty =
(Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Value.weight_quan - wehit_cou) / 2;
if (Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Value.bursty > 0)
{
l_s.Add(Datas.frequency[Datas.now_time_read % 6].ElementAt(i).Key);
l_i.Add(i);
}
}
//距离聚类
l_julei = new List<ju_word>();
for (i = 0; i < l_s.Count; i++)
{
ju_word ju_t = new ju_word(l_s.ElementAt(i), Datas.frequency[Datas.now_time_read % 6][l_s.ElementAt(i)].position);
l_julei.Add(ju_t);
}
double f_z = 0;
double ju_temp;
double ju_min;
double fengzi = 0;
double fenmu = 0;
int temp_count;//the nunmber of the same
int ju_x = 0;
int ju_y = 0;
List<int> l_posi = new List<int>();
while (f_z < 15)
{
ju_min = 10000;
ju_x = 0;
ju_y = 0;
List<int> l_temp_pos = new List<int>();
for (i = 0; i < l_julei.Count; i++)
{
for (j = i + 1; j < l_julei.Count; j++)
{
//zuo = l_julei.ElementAt(i).Split(',');
//you = l_julei.ElementAt(j).Split(',');
for (int x = 0; x < l_julei.ElementAt(i).word.Count; x++)
{
for (int y = 0; y < l_julei.ElementAt(j).word.Count; y++)
{
fengzi += (Datas.frequency[Datas.now_time_read % 6][l_julei.ElementAt(i).word.ElementAt(x)].bursty
* Datas.frequency[Datas.now_time_read % 6][l_julei.ElementAt(j).word.ElementAt(y)].bursty);
}
}
fengzi = fengzi / (l_julei.ElementAt(i).word.Count + l_julei.ElementAt(i).word.Count);
temp_count = 0;
for (int ju_i = 0; ju_i < l_julei.ElementAt(i).position.Count; ju_i++)
{
if (l_julei.ElementAt(j).position.Contains(l_julei.ElementAt(i).position.ElementAt(ju_i)))
{
temp_count++;
l_temp_pos.Add(l_julei.ElementAt(i).position.ElementAt(ju_i));
}
}
fenmu = (double)temp_count / (double)Datas.now_time_max;
ju_temp = fengzi / fenmu;
if (ju_temp < ju_min)
{
ju_min = ju_temp;
ju_x = i;
ju_y = j;
l_posi = new List<int>(l_temp_pos);
}
fengzi = 0;
fenmu = 0;
l_temp_pos.Clear();
}
te = l_julei.OrderBy(ju_word => ju_word.word.Count).ToList();
}
if (ju_x == 0 && ju_y == 0)
break;
//聚类一个
for (i = 0; i < l_julei.ElementAt(ju_y).word.Count; i++)
l_julei.ElementAt(ju_x).word.Add(l_julei.ElementAt(ju_y).word.ElementAt(i));
l_julei.ElementAt(ju_x).position = new List<int>(l_posi);
l_julei.RemoveAt(ju_y);
f_z = ju_min;
}
List<ju_word> temp = new List<ju_word>();
for (i = 0; i < l_julei.Count; i++)
if (l_julei.ElementAt(i).word.Count > 1)
temp.Add(l_julei.ElementAt(i));
Datas.ju_list[Datas.now_time_read%6] = temp.OrderByDescending(ju_word => ju_word.position.Count).ToList();
//this.Invoke(new updateListboxFreq(doUpdateListboxFreq), new object[] { Datas.ju_list });
if (!Datas.ProgressBarCompelet)
Datas.ProgressBarCompelet = true;
}
Datas.now_time_count = 0;
Datas.now_time_read++;
}
}
}
}
