public List <string> print_highwordcount()
{
classif_fn = Select_text.ReadFN();
ver.dictdata(classif_fn, fn);
foreach (string k in ver.dict2.Keys)
{
string tempstr = "";
ver.dict2.TryGetValue(k, out tempstr);
string[] tempstrarr = tempstr.Split('|');
templist.Add(tempstr[0] + " - " + k);
}
templist.Sort();
templist.Reverse();
int count = 0;
foreach (string f in templist)
{
if ((!list_highwordcountstem.Contains(ws.Stem(f))) && count < 25)
{
list_highwordcountstem.Add(ws.Stem(f));
list_highwordcount.Add(f);
count++;
}
}
return(list_highwordcount);
}
public List <string> printword_ver_TFSLF()
{
classif_fn = Select_text.ReadFN();
strlist = tfidf.TF_SLF_ver(classif_fn, fn, ver.dict2, ver.count_all_world);
strlist.Sort();
return(strlist);
}
public String printorigintext()
{
origintext = "";
classif_fn = Select_text.ReadFN();
ver.word_list(classif_fn, fn);
origintext = ver.origintext;
return(origintext);
}
public List <string> printwordsw()
{
Clearlistdict();
classif_fn = Select_text.ReadFN();
ver.word_list(classif_fn, fn);
list_word_sw = ver.lwsw;
return(list_word_sw);
}
public List <string> printstr()
{
Clearlistdict();
classif_fn = Select_text.ReadFN();
ver.word_list(classif_fn, fn);
list_str = ver.lstr;
/*for (int i = 0; i < list_str.Count;i++)
* if(list_str[i].Trim() != "")
* list_str[i] = FirstUpper(list_str[i].Trim());*/
return(list_str);
}
public List <string> printword_count()
{
Clearlistdict();
classif_fn = Select_text.ReadFN();
ver.dictdata(classif_fn, fn);
foreach (string k in ver.dict2.Keys)
{
string tempstr = "";
ver.dict2.TryGetValue(k, out tempstr);
list.Add(k + " - " + tempstr);
}
list.Add("All word count = " + ver.count_all_world.ToString());
return(list);
}
public float [] xdata()
{
string[] xdataarr;
float[] xdataarrfloat;
if (!File.Exists("model.txt"))
{
throw new FileNotFoundException();
}
string data = File.ReadAllText("model.txt", Encoding.Default);
string fn = Select_text.ReadFN();
if (!File.Exists(fn))
{
throw new FileNotFoundException();
}
datatosum = File.ReadAllText(fn, Encoding.Default);
string[] arr_data = data.Split('\n');
arr_datatosum = datatosum.Split('.', '!', '?');
string[] arr_datatosum_word = datatosum.Split(' ');
string[] arr_data_wordsvec = data.Split(' ');
string[] arr_data_words = new string[arr_data_wordsvec.Length / 101 + 1];
int j = 0;
for (int i = 0; i < arr_data_wordsvec.Length; i += 101)
{
arr_data_words[j] = arr_data_wordsvec[i];
j++;
}
j = 0;
int k = 0, flag = 0, kk = 0;
xdataarr = new string[arr_datatosum_word.Length * 100];
for (int i = 0; i < arr_datatosum_word.Length; i++)
{
for (int jj = 0; jj < arr_data_words.Length; jj++)
{
string[] arr_data_item = arr_data_words[jj].Split('\n');
if (arr_datatosum_word[i] == arr_data_item[arr_data_item.Length - 1])
{
for (int u = jj * 101 + 1; u < jj * 101 + 101; u++) //j
{
if (u < arr_data_wordsvec.Length)
{
if (arr_data_wordsvec[u].Trim() != "")
{
xdataarr[k] = arr_data_wordsvec[u].Trim();
}
k++;
flag = 1;
}
}
j = 0;
}
j++;
if (jj == arr_data_words.Length - 1) //j
{
j = 0;
if (flag == 0)
{
kk = k;
for (int u = kk; u < kk + 100; u++)
{
xdataarr[u] = null;
k++;
}
}
flag = 0;
}
}
}
xdataarrfloat = converttofloat(xdataarr);
return(xdataarrfloat);
}