// counts the amount of times a attribute is used in the workload queries
static void freqCounter(string v, string k, int times)
{
v = trimString(v);
if (double.TryParse(v, out double test))
{
return;
}
tuple <string, string> temp = new tuple <string, string>(k, v);
if (freqDict.ContainsKey(temp))
{
freqDict[temp] += times;
}
else
{
freqDict.Add(temp, times);
}
if (maxDict.ContainsKey(k))
{
if (maxDict[k] < freqDict[temp])
{
maxDict[k] = freqDict[temp];
}
}
else
{
maxDict.Add(k, freqDict[temp]);
}
}
// calculates the simulator of a tuple and a query
static tuple <int, double>[] calculateTopK(string[] columns, string[] values, int k)
{
string sql = "select * from autompg";
SQLiteCommand com = new SQLiteCommand(sql, m_dbConnection);
SQLiteDataReader reader = com.ExecuteReader();
tuple <int, double>[] topKTuples = new tuple <int, double> [k];
for (int i = 0; i < k; i++)
{
topKTuples[i] = new tuple <int, double>(-1, 0);
}
Dictionary <int, double> MissingAttributesValues = new Dictionary <int, double>();
while (reader.Read())
{
double simScore = 0;
for (int i = 0; i < values.Length; i++)
{
string sqlMeta = "select idfqf from idfqf where column = '" + columns[i] + "' and value = " + values[i];
SQLiteCommand metaCom = new SQLiteCommand(sqlMeta, meta_dbConnection);
SQLiteDataReader metaReader = metaCom.ExecuteReader();
if (metaReader.Read())
{
// numerical similarity
if (double.TryParse(values[i], out double q))
{
// t value van de tuple, q value van de query
double innerproduct = ((double.Parse(reader[columns[i]].ToString()) - q) / hColumnDict[columns[i]]);
simScore += Math.Exp(-0.5 * innerproduct * innerproduct) * (double)metaReader.GetValue(0);
}
// categorical similarity
else
{
string sqlJaq = "select jacq from jacquard where value_1 = " + values[i] + " and value_2 = '" + reader[columns[i]] + "'";
SQLiteCommand jaqCom = new SQLiteCommand(sqlJaq, meta_dbConnection);
SQLiteDataReader jaqReader = jaqCom.ExecuteReader();
if (jaqReader.Read())
{
simScore += (double)jaqReader.GetValue(0) * (double)metaReader.GetValue(0);
}
}
}
else if (double.TryParse(values[i], out double q))
{
double denom = 0;
sql = "select " + columns[i] + " from autompg";
SQLiteCommand idfcom = new SQLiteCommand(sql, m_dbConnection);
SQLiteDataReader r = idfcom.ExecuteReader();
while (r.Read())
{
double ti = double.Parse(r[columns[i]].ToString());
double diff = ti - q;
denom += Math.Exp(-0.5 * (diff / hColumnDict[columns[i]]) * (diff / hColumnDict[columns[i]]));
}
double idfnumber = Math.Log10(N / denom);
double innerproduct = ((double.Parse(reader[columns[i]].ToString()) - q) / hColumnDict[columns[i]]);
simScore += Math.Exp(-0.5 * innerproduct * innerproduct) * idfnumber;
}
}
double missingscore = 0;
// calculate missing score
for (int i = 0; i < categoricalNames.Length; i++)
{
string sqlMeta = "select idfqf from idfqf where column = '" + categoricalNames[i] + "' and value = '" + reader[categoricalNames[i]] + "'";
SQLiteCommand metaCom = new SQLiteCommand(sqlMeta, meta_dbConnection);
SQLiteDataReader metaReader = metaCom.ExecuteReader();
if (metaReader.Read())
{
missingscore += Math.Log10((double)metaReader.GetValue(0));
}
}
MissingAttributesValues.Add(int.Parse(reader["id"].ToString()), missingscore);
//calculate top k using similarity score
for (int i = 0; i < k; i++)
{
if (topKTuples[i].first == -1)
{
topKTuples[i] = new tuple <int, double>(int.Parse(reader["id"].ToString()), simScore);
break;
}
if (topKTuples[i].second == simScore)
{
if (MissingAttributesValues[topKTuples[i].first] < MissingAttributesValues[int.Parse(reader["id"].ToString())])
{
for (int j = k - 1; j > i; j--)
{
topKTuples[j] = topKTuples[j - 1];
}
topKTuples[i] = new tuple <int, double>(int.Parse(reader["id"].ToString()), simScore);
break;
}
}
if (topKTuples[i].second < simScore)
{
for (int j = k - 1; j > i; j--)
{
topKTuples[j] = topKTuples[j - 1];
}
topKTuples[i] = new tuple <int, double>(int.Parse(reader["id"].ToString()), simScore);
break;
}
}
}
return(topKTuples);
}
