private string GenerateAdjacencyMatrix()
{
int n = _pdg.Nodes.Count();
double defaultValue = 1.0 / n;
double[] rowValues = new double[n];
FileStream matrixFS = RUtil.CreateCacheFile("HITS." + _traceID + ".TPM.matrix");
TextWriter matrixWriter = new StreamWriter(matrixFS);
//FileStream edgeFS = RUtil.CreateCacheFile("HITS." + _traceID + ".TPM.edges");
//TextWriter edgeWriter = new StreamWriter(edgeFS);
FileStream mapFS = RUtil.CreateCacheFile("HITS." + _traceID + ".TPM.map");
_mappingFile = mapFS.Name;
TextWriter mapWriter = new StreamWriter(mapFS);
for (int nodeIndex = 0; nodeIndex < _pdg.Nodes.Count(); nodeIndex++)
{
PDGNode pdgNode = _pdg.GetNode(nodeIndex);
for (int i = 0; i < n; i++)
{
rowValues[i] = 0;
}
//edgeWriter.WriteLine(pdgNode.OutgoingEdges.Count()); // write number of outgoing edges for Topical HITS algorithm
for (int indexOutgoingEdge = 0; indexOutgoingEdge < pdgNode.OutgoingEdges.Count(); indexOutgoingEdge++)
{
PDGEdge pdgOutgoingEdge = pdgNode.OutgoingEdges.ElementAt(indexOutgoingEdge);
int columnFrequencies = _pdg.IndexOf(pdgOutgoingEdge.OutgoingNodeID);
// for positive values only
if ((columnFrequencies < 0))
{
throw new RDataException();
// continue;
}
if (_config.Weight == WebMiningWeightEnum.Binary)
{
rowValues[columnFrequencies] = 1;
}
else
if (_config.Weight == WebMiningWeightEnum.Frequency)
{
rowValues[columnFrequencies] = pdgOutgoingEdge.Weight;
}
else
{
throw new RDataException("Unknown weighting scheme: " + _config.Weight);
}
}
//for (int i=1;i<=n;i++)
//{
// matrixWriter.Write(rowValuesFrequencies[i]+" ");
// binaryWriter.Write(rowValuesBinary[i]+" ");
//}
matrixWriter.WriteLine(String.Join(" ", rowValues));
mapWriter.WriteLine(pdgNode.MethodName);
}
matrixWriter.Flush();
matrixWriter.Close();
//edgeWriter.Flush();
//edgeWriter.Close();
mapWriter.Flush();
mapWriter.Close();
return(matrixFS.Name);
}
private string GenerateTransitionProbabilityMatrix()
{
int n = _pdg.Nodes.Count();
double defaultValue = 1.0 / n;
double[] rowValues = new double[n];
FileStream matrixFS = RUtil.CreateCacheFile("PageRank." + _traceID + ".TPM.matrix");
TextWriter matrixWriter = new StreamWriter(matrixFS);
FileStream edgeFS = RUtil.CreateCacheFile("PageRank." + _traceID + ".TPM.edges");
TextWriter edgeWriter = new StreamWriter(edgeFS);
FileStream mapFS = RUtil.CreateCacheFile("PageRank." + _traceID + ".TPM.map");
_mappingFile = mapFS.Name;
TextWriter mapWriter = new StreamWriter(mapFS);
for (int nodeIndex = 0; nodeIndex < _pdg.Nodes.Count(); nodeIndex++)
{
PDGNode pdgNode = _pdg.GetNode(nodeIndex);
if (pdgNode.OutgoingEdges.Count() == 0)
{
for (int i = 0; i < n; i++)
{
rowValues[i] = defaultValue;
}
}
else
{
for (int i = 0; i < n; i++)
{
rowValues[i] = 0.0;
}
}
edgeWriter.WriteLine(pdgNode.OutgoingEdges.Count()); // write number of outgoing edges for most of the advanced PageRank algorithms
for (int indexOutgoingEdge = 0; indexOutgoingEdge < pdgNode.OutgoingEdges.Count(); indexOutgoingEdge++)
{
PDGEdge pdgOutgoingEdge = pdgNode.OutgoingEdges.ElementAt(indexOutgoingEdge);
int columnFrequencies = _pdg.IndexOf(pdgOutgoingEdge.OutgoingNodeID);
// for positive values only
if (columnFrequencies == -1)
{
throw new RDataException("Invalid column index.");
// continue;
}
rowValues[columnFrequencies] = pdgOutgoingEdge.Weight;
}
//for (int i=1;i<=n;i++)
//{
// matrixWriter.Write(rowValues[i]+" ");
//}
matrixWriter.WriteLine(String.Join(" ", rowValues));
mapWriter.WriteLine(pdgNode.MethodName);
}
matrixWriter.Flush();
matrixWriter.Close();
edgeWriter.Flush();
edgeWriter.Close();
mapWriter.Flush();
mapWriter.Close();
return(matrixFS.Name);
}
