static void TestPathMining()
{
IndexedGraph g = new IndexedGraph();
g.Read(@"D:\Users\v-jiahan\HORM\Data\ex3_graph.txt");
FrequentPathMining fpm = new FrequentPathMiningBreadth();
fpm.Init(g, 1000, 3, true);
ShowingNames showname = new ShowingNames(@"D:\Users\v-jiahan\HORM\Data\ex3_graph");
ShowResult(fpm._resultCache, showname);
}
//public List<Tuple<IndexedGraph, int>> Mine(int minSupp, int maxSize, int[] constraintVSet)
//{
// DateTime begin = DateTime.Now;
// //FrequentPathMining fpm = new FrequentPathMiningDepth();
// FrequentPathMining fpm = new FrequentPathMiningBreadth();
// fpm.Init(_g, minSupp, maxSize, constraintVSet);
// //fpm.Init(_g, minSupp, maxSize);
// Console.WriteLine("{0} seconds. {1} path results.",
// (DateTime.Now - begin).TotalSeconds,
// fpm._resultCache.Count);
// List<Tuple<IndexedGraph, int>> ret = new List<Tuple<IndexedGraph, int>>();
// Console.WriteLine("Adding {0} paths.", fpm.GetPath(1).Count);
// List<IndexedGraph> lastResults = fpm.GetPath(1);
// ret.AddRange(fpm.GetPathAndSupp(1));
// for (int size = 2; size <= maxSize; size++)
// {
// begin = DateTime.Now;
// Console.WriteLine("Computing Size-{0} Candidate Patterns.", size);
// List<IndexedGraph> tempResults = new List<IndexedGraph>();
// for (int i = 0; i < lastResults.Count; i++)
// {
// for (int j = i; j < lastResults.Count; j++)
// {
// List<IndexedGraph> newPatterns = null;
// if (j == i)
// {
// Graph tempG = lastResults[i].ShallowCopy();
// newPatterns = JoinGraphPair(tempG, lastResults[i]);
// }
// else
// newPatterns = JoinGraphPair(lastResults[i], lastResults[j]);
// foreach (IndexedGraph pattern in newPatterns)
// {
// bool hasDup = false;
// foreach (IndexedGraph pattern1 in tempResults)
// if (_subGTester.MatchNeighborhood(pattern, pattern1, 0))
// {
// hasDup = true;
// break;
// }
// if (!hasDup)
// tempResults.Add(pattern);
// }
// }
// }
// Console.WriteLine("{0} seconds. {1} candidates.",
// (DateTime.Now - begin).TotalSeconds,
// tempResults.Count);
// begin = DateTime.Now;
// Console.WriteLine("Validating Size-{0} Candidate Patterns.", size);
// lastResults.Clear();
// foreach (IndexedGraph pattern in tempResults)
// {
// int supp = 0;
// List<List<int>> ccands = _niindex.GetCandidatesForeachVertex(pattern);
// if (ccands.Exists(e => e != null && e.Count == 0))
// continue;
// if (ccands[0].Count < minSupp)
// continue;
// _subGTester._cands = ccands;
// foreach (int i in constraintVSet)
// if (_subGTester.MatchNeighborhood(pattern, _g, i))
// supp++;
// _subGTester._cands = null;
// if (supp >= minSupp)
// {
// ret.Add(new Tuple<IndexedGraph, int>(pattern, supp));
// //pattern.Print();
// //Console.WriteLine(supp+"\n");
// lastResults.Add(pattern);
// }
// }
// Console.WriteLine("{0} seconds. {1} results.",
// (DateTime.Now - begin).TotalSeconds,
// lastResults.Count);
// begin = DateTime.Now;
// lastResults.AddRange(fpm.GetPath(size));
// Console.WriteLine("Adding {0} paths.", fpm.GetPath(size).Count);
// //lastResults.Sort((x, y) => x._vertexes.Length - y._vertexes.Length);
// ret.AddRange(fpm.GetPathAndSupp(size));
// if (lastResults.Count == 0)
// break;
// }
// return ret;
//}
public List <Tuple <IndexedGraph, int> > Mine(int minSupp, int maxSize, int[] constraintVSet)
{
bool useVIDList = true;
List <int> constraintVSetList = constraintVSet.ToList();
DateTime begin = DateTime.Now;
//FrequentPathMining fpm = new FrequentPathMiningDepth();
FrequentPathMining fpm = new FrequentPathMiningBreadth();
fpm.Init(_g, minSupp, maxSize, constraintVSet, useVIDList);
//fpm.Init(_g, minSupp, maxSize);
Console.WriteLine("{0} seconds. {1} path results.",
(DateTime.Now - begin).TotalSeconds,
fpm._resultCache.Count);
List <Tuple <IndexedGraph, int> > ret = new List <Tuple <IndexedGraph, int> >();
Console.WriteLine("Adding {0} paths.", fpm.GetPathAndInfo(1).Count);
List <Tuple <IndexedGraph, List <int> > > lastResults = fpm.GetPathAndVID(1);
ret.AddRange(fpm.GetPathAndCount(1));
for (int size = 2; size <= maxSize; size++)
{
begin = DateTime.Now;
Console.WriteLine("Computing Size-{0} Candidate Patterns.", size);
List <Tuple <IndexedGraph, List <int> > > tempResults = new List <Tuple <IndexedGraph, List <int> > >();
for (int i = 0; i < lastResults.Count; i++)
{
for (int j = i; j < lastResults.Count; j++)
{
List <IndexedGraph> newPatterns = null;
List <int> vids = null;
if (j == i)
{
Graph tempG = lastResults[i].Item1.ShallowCopy();
newPatterns = JoinGraphPair(tempG, lastResults[i].Item1);
vids = lastResults[i].Item2;
}
else
{
if (useVIDList)
{
vids = Tools.MergeSortedArray(lastResults[i].Item2, lastResults[j].Item2);
if (vids.Count < minSupp)
{
continue;
}
}
newPatterns = JoinGraphPair(lastResults[i].Item1, lastResults[j].Item1);
}
foreach (IndexedGraph pattern in newPatterns)
{
bool hasDup = false;
foreach (var pair in tempResults)
{
if (_subGTester.MatchNeighborhood(pattern, pair.Item1, 0))
{
hasDup = true;
break;
}
}
if (!hasDup)
{
tempResults.Add(new Tuple <IndexedGraph, List <int> >(pattern, vids));
}
}
}
}
Console.WriteLine("{0} seconds. {1} candidates.",
(DateTime.Now - begin).TotalSeconds,
tempResults.Count);
begin = DateTime.Now;
Console.WriteLine("Validating Size-{0} Candidate Patterns.", size);
lastResults.Clear();
foreach (var pair in tempResults)
{
int supp = 0;
List <List <int> > ccands = _niindex.GetCandidatesForeachVertex(pair.Item1);
if (ccands.Exists(e => e != null && e.Count == 0))
{
continue;
}
if (ccands[0].Count < minSupp)
{
continue;
}
_subGTester._cands = ccands;
List <int> vids = null, filteredVids = null;
if (useVIDList)
{
filteredVids = new List <int>();
vids = pair.Item2;
}
else
{
vids = constraintVSetList;
}
foreach (int i in vids)
{
if (_subGTester.MatchNeighborhood(pair.Item1, _g, i))
{
supp++;
if (useVIDList)
{
filteredVids.Add(i);
}
}
}
_subGTester._cands = null;
if (supp >= minSupp)
{
ret.Add(new Tuple <IndexedGraph, int>(pair.Item1, supp));
//pattern.Print();
//Console.WriteLine(supp+"\n");
lastResults.Add(new Tuple <IndexedGraph, List <int> >(pair.Item1, filteredVids));
}
}
Console.WriteLine("{0} seconds. {1} results.",
(DateTime.Now - begin).TotalSeconds,
lastResults.Count);
begin = DateTime.Now;
var addpath = fpm.GetPathAndVID(size);
lastResults.AddRange(addpath);
Console.WriteLine("Adding {0} paths.", addpath.Count);
//lastResults.Sort((x, y) => x._vertexes.Length - y._vertexes.Length);
ret.AddRange(fpm.GetPathAndCount(size));
if (lastResults.Count == 0)
{
break;
}
}
return(ret);
}
public List <Tuple <IndexedGraph, List <int> > > MineEgonet(int minSupp, int maxSize, int maxRadius, int[] constraintVSet)
{
//bool useVIDList = false;
List <int> constraintVSetList = constraintVSet.ToList();
DateTime begin = DateTime.Now;
//FrequentPathMining fpm = new FrequentPathMiningDepth();
FrequentPathMining fpm = new FrequentPathMiningBreadth();
fpm.Init(_g, minSupp, maxSize, constraintVSet, true, maxRadius);
//fpm.Init(_g, minSupp, maxSize);
Console.WriteLine("{0} seconds. {1} path results.",
(DateTime.Now - begin).TotalSeconds,
fpm._resultCache.Count);
List <Tuple <IndexedGraph, List <int> > > ret = new List <Tuple <IndexedGraph, List <int> > >();
Console.WriteLine("Adding {0} paths.", fpm.GetPathAndVID(1).Count);
List <Tuple <IndexedGraph, List <int> > > lastResults = fpm.GetPathAndVID(1);
ret.AddRange(fpm.GetPathAndVID(1));
//IndexedGraph z = new IndexedGraph();
//z.Read(@"E:\RkNPQ\1.txt");
//SubGraphTest sgt1 = new SubGraphTest();
for (int size = 2; size <= maxSize; size++)
{
begin = DateTime.Now;
Console.WriteLine("Computing Size-{0} Candidate Patterns.", size);
List <Tuple <IndexedGraph, List <int> > > tempResults = new List <Tuple <IndexedGraph, List <int> > >();
for (int i = 0; i < lastResults.Count; i++)
{
for (int j = i; j < lastResults.Count; j++)
{
List <IndexedGraph> newPatterns = null;
List <int> vids = null;
if (j == i)
{
Graph tempG = lastResults[i].Item1.ShallowCopy();
newPatterns = JoinGraphPair(tempG, lastResults[i].Item1);
vids = lastResults[i].Item2;
}
else
{
vids = Tools.MergeSortedArray(lastResults[i].Item2, lastResults[j].Item2);
if (vids.Count < minSupp)
{
continue;
}
newPatterns = JoinGraphPair(lastResults[i].Item1, lastResults[j].Item1);
}
foreach (IndexedGraph pattern in newPatterns)
{
//if (sgt1.MatchNeighborhood(z, pattern, 0))
//{
// Console.WriteLine();
//}
bool hasDup = false;
foreach (var pair in tempResults)
{
if (_subGTester.MatchNeighborhood(pattern, pair.Item1, 0))
{
hasDup = true;
break;
}
}
if (!hasDup)
{
tempResults.Add(new Tuple <IndexedGraph, List <int> >(pattern, vids));
}
}
}
}
Console.WriteLine("{0} seconds. {1} candidates.",
(DateTime.Now - begin).TotalSeconds,
tempResults.Count);
begin = DateTime.Now;
Console.WriteLine("Validating Size-{0} Candidate Patterns.", size);
//Compute Zipper Patterns
List <Tuple <IndexedGraph, List <int> > > zipperPatterns = new List <Tuple <IndexedGraph, List <int> > >();
if (size > maxRadius + 1 && size <= 2 * maxRadius + 1)
{
SubGraphTest sgt = new SubGraphTest(true, MapOperationInstances.GetZipperHeads);
foreach (var tup in lastResults)
{
//if(sgt1.MatchNeighborhood(z,tup.Item1,0))
// tup.Item1.Print();
var headPosList = ZipperHandler.DetectPotentialZipper(tup.Item1, maxRadius);
if (headPosList == null)
{
continue;
}
List <int> vids = null;
Dictionary <Tuple <int, int>, List <int> > mapELabel2Vids = new Dictionary <Tuple <int, int>, List <int> >();
vids = tup.Item2;
foreach (int i in vids)
{
sgt.MatchNeighborhood(tup.Item1, _g, i, headPosList);
HashSet <Tuple <int, int> > eLabels = new HashSet <Tuple <int, int> >();
sgt._rets.ForEach(e => eLabels.Add((Tuple <int, int>)e));
foreach (var pair in eLabels)
{
if (!mapELabel2Vids.ContainsKey(pair))
{
mapELabel2Vids[pair] = new List <int>();
}
mapELabel2Vids[pair].Add(i);
}
}
foreach (var pair in mapELabel2Vids)
{
if (pair.Value.Count >= minSupp)
{
Edge newEdge = new Edge();
newEdge._from = headPosList[pair.Key.Item2].Item1;
newEdge._to = headPosList[pair.Key.Item2].Item2;
newEdge._eLabel = pair.Key.Item1;
IndexedGraph g = tup.Item1.ShallowCopy();
Edge[] newEdgeList = new Edge[g._edges.Length + 1];
g._edges.CopyTo(newEdgeList, 0);
newEdgeList[newEdgeList.Length - 1] = newEdge;
g._edges = newEdgeList;
Vertex v1 = g._vertexes[newEdge._from];
int[] newOutEdgeList = new int[v1._outEdge.Length + 1];
v1._outEdge.CopyTo(newOutEdgeList, 0);
newOutEdgeList[newOutEdgeList.Length - 1] = newEdgeList.Length - 1;
v1._outEdge = newOutEdgeList;
Vertex v2 = g._vertexes[newEdge._to];
int[] newInEdgeList = new int[v2._inEdge.Length + 1];
v2._inEdge.CopyTo(newInEdgeList, 0);
newInEdgeList[newInEdgeList.Length - 1] = newEdgeList.Length - 1;
v2._inEdge = newInEdgeList;
g.GenIndex();
zipperPatterns.Add(new Tuple <IndexedGraph, List <int> >(g, pair.Value));
}
}
}
}
lastResults.Clear();
foreach (var pair in tempResults)
{
int supp = 0;
List <List <int> > ccands = _niindex.GetCandidatesForeachVertex(pair.Item1);
if (ccands.Exists(e => e != null && e.Count == 0))
{
continue;
}
//if (ccands[0].Count < minSupp)
//continue;
_subGTester._cands = ccands;
List <int> vids = null, filteredVids = null;
filteredVids = new List <int>();
vids = pair.Item2;
foreach (int i in vids)
{
if (_subGTester.MatchNeighborhood(pair.Item1, _g, i))
{
supp++;
filteredVids.Add(i);
}
}
_subGTester._cands = null;
if (supp >= minSupp)
{
ret.Add(new Tuple <IndexedGraph, List <int> >(pair.Item1, filteredVids));
//pattern.Print();
//Console.WriteLine(supp+"\n");
lastResults.Add(new Tuple <IndexedGraph, List <int> >(pair.Item1, filteredVids));
}
}
Console.WriteLine("{0} seconds. {1} results.",
(DateTime.Now - begin).TotalSeconds,
lastResults.Count);
begin = DateTime.Now;
if (size <= maxRadius + 1)
{
var addpath = fpm.GetPathAndVID(size);
lastResults.AddRange(addpath);
Console.WriteLine("Adding {0} paths.", addpath.Count);
//lastResults.Sort((x, y) => x._vertexes.Length - y._vertexes.Length);
ret.AddRange(fpm.GetPathAndVID(size));
}
else
{
//add Zipper Patterns
Console.WriteLine("Adding {0} Zippers.", zipperPatterns.Count);
lastResults.AddRange(zipperPatterns);
ret.AddRange(zipperPatterns);
}
if (lastResults.Count == 0)
{
break;
}
}
return(ret);
}
