public List <Rule> StartProcess(double minSupport, CancellationToken token = new CancellationToken())
{
rules.Clear();
this.token = token;
this.minSupport = minSupport * dataset.countRows;
var tidList = new Dictionary <BitSet, HashSet <int> >();
foreach (var item in dataset.tidList)
{
if (item.Value.Count >= minSupport)
{
tidList.Add(item.Key, item.Value);
}
}
tidArr = new BitSet[tidList.Count];
tidHashArr = new HashSet <int> [tidList.Count];
int itemTid = 0;
foreach (var item in tidList)
{
tidArr[itemTid] = item.Key;
tidHashArr[itemTid] = item.Value;
itemTid++;
}
Parallel.For(0, tidArr.Length,
(item, st) =>
{
Process(tidArr[item], tidHashArr[item], item);
if (this.token.IsCancellationRequested)
{
st.Break();
}
ProgressReportEvent?.Invoke(this, new ProgressChangedArgs(1));
});
var list = new List <Rule>();
foreach (var item in rules)
{
list.Add(new Rule(item.Key, item.Value / countRows));
}
return(list);
}
private void Process()
{
var toAddDict = new Dictionary <BitSet, HashSet <int> >();
do
{
toAddDict.Clear();
var top = tidList.GetEnumerator();
for (int i = 0; i < tidList.Count; i++)
{
top.MoveNext();
var item = top.Current;
Debug.Assert(item.Key != null, "Top out");
var key = new BitSet(countColumns);
var inner = tidList.GetEnumerator();
for (int j = 0; j < i; j++)
{
inner.MoveNext();
}
for (int j = i; j < tidList.Count; j++)
{
inner.MoveNext();
var value = inner.Current;
Debug.Assert(value.Key != null, "Inner out");
key.SetFrom(item.Key);
key.Or(value.Key);
if (key.Equals(item.Key) ||
key.Equals(value.Key) ||
tidList.ContainsKey(key) ||
toAddDict.ContainsKey(key))
{
continue;
}
var hashSet = new HashSet <int>(item.Value);
hashSet.IntersectWith(value.Value);
double support = hashSet.Count;
if (support < minSupport)
{
continue;
}
var keyNew = new BitSet(key);
toAddDict.Add(keyNew, hashSet);
rules.Add(new Rule(keyNew, support));
}
}
foreach (var item in toAddDict)
{
tidList.Add(item.Key, item.Value);
}
if (token.IsCancellationRequested)
{
return;
}
ProgressReportEvent?.Invoke(this, new ProgressChangedArgs(1));
} while (toAddDict.Count != 0);
}