static void Main(string[] args)
{
ClassLabelDictionary classLabelDict = new ClassLabelDictionary();
classLabelDict.LoadFromDB();
TFIDFDictionary tfidfDict = new TFIDFDictionary();
tfidfDict.LoadFromDB();
PrecedenceModel precedenceModel = new PrecedenceModel(tfidfDict, classLabelDict);
precedenceModel.DiscoverPrecedence();
}
public void GenerateTFIDFDictionary()
{
Dictionary<int, TFIDF> tfidfDict = new Dictionary<int, TFIDF>();
foreach (DocModel m in docDB)
{
for (int i = 0; i < m.Length; i++)
{
int wordKey = m.Word(i);
TFIDF tfidf;
if (!tfidfDict.TryGetValue(wordKey, out tfidf))
{
tfidf = new TFIDF(wordKey, docDB.Count);
tfidf.key = wordKey;
tfidfDict.Add(wordKey, tfidf);
}
tfidf.tf += m.Count(i);
tfidf.df++;
}
}
List<KeyValuePair<int, TFIDF>> tfidfs = tfidfDict.ToList();
tfidfs.Sort(
(x1, x2) =>
{
if (x1.Value.tfidf > x2.Value.tfidf)
{
return -1;
}
else if (x1.Value.tfidf == x2.Value.tfidf)
{
return 0;
}
else
{
return 1;
}
});
TFIDFDictionary tfidfDictionary = new TFIDFDictionary();
for (int i = 0; i < tfidfs.Count && i < 20000; i++)
{
tfidfDictionary.AddValue(wordDict.GetKey(tfidfs[i].Key), tfidfs[i].Key);
}
tfidfDictionary.StoreToDB();
}
static void Main(string[] args)
{
ClassLabelDictionary classLabelDict = new ClassLabelDictionary();
classLabelDict.LoadFromDB();
WordDictionary wordDict = new WordDictionary();
wordDict.LoadFromDB();
TFIDFDictionary tfidfDict = new TFIDFDictionary();
tfidfDict.LoadFromDB();
BoWModelDB docDB = new BoWModelDB(wordDict);
docDB.LoadFromDBByDataSet("doc_set_cls_1000");
PModel.PrecedenceModel pModel = new PModel.PrecedenceModel(tfidfDict, classLabelDict);
pModel.DiscoverPrecedence();
PrecedenceQuery pQuery = new PrecedenceQuery(pModel, wordDict, classLabelDict);
pQuery.TestQuery(docDB);
}
static void InitTFIDFDict()
{
classLabelDict = new ClassLabelDictionary();
classLabelDict.LoadFromDB();
tfidfDict = new TFIDFDictionary();
tfidfDict.LoadFromDB();
}