public Hashtable GetVector()
{
StringBuilder sb = new StringBuilder();
string indexPath = AppDomain.CurrentDomain.BaseDirectory.ToString() + "\\App_Data\\";
DirectoryInfo dirInfo = new DirectoryInfo(indexPath);
FSDirectory dir = FSDirectory.Open(dirInfo);
Hashtable ht = new Hashtable();
Hashtable ht2 = new Hashtable();
IndexReader ir = IndexReader.Open(dir, false);
DefaultSimilarity similarity = new DefaultSimilarity();
string[] label = null;
int[] freq = null;
foreach (var obj in ir.GetTermFreqVectors(0))
{
label = obj.GetTerms();
freq = obj.GetTermFrequencies();
}
Response.Write("s1<br>");
for (int i = 0; i <= label.Length - 1; i++)
{
ht.Add(label[i], freq[i]);
if (!TotalNoun.Contains(label[i])) {
TotalNoun.Add(label[i]);
}
}
foreach (DictionaryEntry obj in ht)
{
float tf = similarity.Tf(Convert.ToInt32(obj.Value));
float wf = TFIDF.WF(tf);
Term t = new Term("Text", obj.Key.ToString());
float idf = similarity.Idf(ir.DocFreq(t), ir.NumDocs());
float wfidf = wf * idf;
sb.AppendFormat(string.Format("label:{0} freq:{1}<BR>", obj.Key, obj.Value));
sb.AppendFormat(string.Format("TF:{0}<br>",similarity.Tf(Convert.ToInt32(obj.Value))));
sb.AppendFormat(string.Format("IDF:{0}<br>", similarity.Idf(ir.DocFreq(t), ir.NumDocs())));
sb.AppendFormat(string.Format("WF-IDF:{0}<br>", wfidf));
ht2.Add(obj.Key, wfidf);
}
Response.Write(sb.ToString());
return ht2;
}
