static void InitAP()
{
wordDict = new WordDictionary();
docModelDB = new LDABoWModelDB(10, wordDict);
string line;
StreamReader reader = new StreamReader(new FileStream("ap.dat", FileMode.Open));
while ((line = reader.ReadLine()) != null)
{
string[] ss = line.Split(' ');
LDABoWModel doc = new LDABoWModel(docModelDB);
for (int i = 1; i < ss.Length; i++)
{
string[] sss = ss[i].Split(':');
doc.AddWord(int.Parse(sss[0]), int.Parse(sss[1]));
wordDict.AddValue(sss[0], int.Parse(sss[0]));
}
doc.InitIndex();
docModelDB.AddDocModel(doc);
}
reader.Close();
docModelDB.Init();
}
static void Stats()
{
InitDict();
docModelDB = new LDABoWModelDB(1, wordDict);
docModelDB.LoadFromDB();
// print class labels
docModelDB.Stats(classLabelDict);
}
static void LDAEstimateDataSet(string dataSetName, int numOfTopics)
{
InitTFIDFDict();
docModelDB = new LDABoWModelDB(numOfTopics, tfidfDict);
docModelDB.LDACollectionName = "ldadocs_" + numOfTopics;
docModelDB.LoadFromDBByDataSet(dataSetName);
docModelDB.Init();
docModelDB.RunEM(true);
docModelDB.SaveLDAModel();
docModelDB = null;
}
static void LDAInferenceDataSet(string dataSetName, int numOfTopics)
{
InitTFIDFDict();
docModelDB = new LDABoWModelDB(numOfTopics, tfidfDict);
docModelDB.LDACollectionName = "ldadocs_" + numOfTopics+ "_all";
docModelDB.LoadFromDBByDataSet(dataSetName);
docModelDB.Init();
docModelDB.LoadLDAModel();
docModelDB.E_Step(true);
docModelDB = null;
}
public LDABoWModel(LDABoWModelDB db)
{
modelDB = db;
Init();
}