/// <summary>
/// Precompute method
/// </summary>
public override void PreCompute()
{
RUtil.RegisterScript(Assembly.GetExecutingAssembly(), _baseScript);
CheckParameters();
_outputFile = RUtil.ReserveCacheFile("LSA.out");
DirectoryInfo corpusDir = SaveArtifactsToCache(_source, _target, "LSA.corpus");
_arguments = new List <object>();
_arguments.Add(corpusDir.FullName);
_arguments.Add(_SourceFile);
_arguments.Add(_TargetFile);
_arguments.Add(_outputFile);
_arguments.Add(_config.Dimensions);
}
private DirectoryInfo SaveArtifactsToCache(TLArtifactsCollection source, TLArtifactsCollection target, string name)
{
DirectoryInfo infoDir = RUtil.CreateCacheDirectory(name);
FileStream sFile = RUtil.CreateCacheFile("LSA.corpus.source");
TextWriter sourceWriter = new StreamWriter(sFile);
_SourceFile = sFile.Name;
FileStream tFile = RUtil.CreateCacheFile("LSA.corpus.target");
TextWriter targetWriter = new StreamWriter(tFile);
_TargetFile = tFile.Name;
FileStream mFile = RUtil.CreateCacheFile("LSA.corpus.map");
TextWriter mapWriter = new StreamWriter(mFile);
_mapFile = mFile.Name;
int fileIndex = 1;
foreach (TLArtifact artifact in source.Values)
{
TextWriter tw = new StreamWriter(Path.Combine(infoDir.FullName, fileIndex.ToString()));
tw.Write(artifact.Text);
tw.Flush();
tw.Close();
sourceWriter.WriteLine(fileIndex);
mapWriter.WriteLine(artifact.Id);
fileIndex++;
}
sourceWriter.Flush();
sourceWriter.Close();
foreach (TLArtifact artifact in target.Values)
{
TextWriter tw = new StreamWriter(Path.Combine(infoDir.FullName, fileIndex.ToString()));
tw.Write(artifact.Text);
tw.Flush();
tw.Close();
targetWriter.WriteLine(fileIndex);
mapWriter.WriteLine(artifact.Id);
fileIndex++;
}
targetWriter.Flush();
targetWriter.Close();
mapWriter.Flush();
mapWriter.Close();
return(infoDir);
}
/// <summary>
/// Precompute method
/// </summary>
public override void PreCompute()
{
RUtil.RegisterScript(Assembly.GetExecutingAssembly(), _baseScript);
DirectoryInfo sourceInfo = SaveArtifactsToCache(_source, "GibbsLDA.source");
DirectoryInfo targetInfo = SaveArtifactsToCache(_target, "GibbsLDA.target");
_outputFile = RUtil.ReserveCacheFile("GibbsLDA.out");
_arguments = new List <object>();
_arguments.Add(sourceInfo.FullName);
_arguments.Add(targetInfo.FullName);
_arguments.Add(_outputFile);
_arguments.Add(_config.NumTopics);
_arguments.Add(_config.GibbsIterations);
_arguments.Add(_config.Alpha);
_arguments.Add(_config.Beta);
_arguments.Add(_config.Seed);
}
/// <summary>
/// Saves corpus to cache.
/// Overwrites existing files with the same name.
/// </summary>
/// <returns>Corpus base path + name</returns>
public LDACorpusInfo Save()
{
LDACorpusInfo info = new LDACorpusInfo();
info.Name = Name;
// write matrix
FileStream cFS = RUtil.CreateCacheFile(Name + ".corpus");
info.Corpus = cFS.Name;
TextWriter corpus = new StreamWriter(cFS);
corpus.Write(Matrix);
corpus.Flush();
corpus.Close();
// write vocab
FileStream vFS = RUtil.CreateCacheFile(Name + ".vocab");
info.Vocab = vFS.Name;
TextWriter vocab = new StreamWriter(vFS);
vocab.Write(Vocab);
vocab.Flush();
vocab.Close();
// write edges
FileStream eFS = RUtil.CreateCacheFile(Name + ".tableWriter");
info.Edges = eFS.Name;
TextWriter edges = new StreamWriter(eFS);
edges.Write(Edges);
edges.Flush();
edges.Close();
// write links
FileStream lFS = RUtil.CreateCacheFile(Name + ".links");
info.Links = lFS.Name;
TextWriter links = new StreamWriter(lFS);
links.Write(Links);
links.Flush();
links.Close();
// return info
return(info);
}
/// <summary>
/// Precompute method
/// </summary>
public override void PreCompute()
{
RUtil.RegisterScript(Assembly.GetExecutingAssembly(), _baseScript);
LDACorpus corpus = new LDACorpus("LDA", _source, _target);
LDACorpusInfo info = corpus.Save();
_outputFile = RUtil.ReserveCacheFile("LDA.out");
_arguments = new List <object>();
_arguments.Add(info.Corpus);
_arguments.Add(info.Vocab);
_arguments.Add(info.Edges);
_arguments.Add(_outputFile);
_arguments.Add(_config.NumTopics);
_arguments.Add(_config.NumIterations);
_arguments.Add(_config.Alpha);
_arguments.Add(_config.Eta);
_arguments.Add(_config.PredictionBeta);
_arguments.Add(_config.Seed);
}
/// <summary>
/// Precompute method
/// </summary>
public override void PreCompute()
{
RUtil.RegisterScript(Assembly.GetExecutingAssembly(), _baseScript);
_info = _corpus.Save();
_outputFile = RUtil.ReserveCacheFile("RTM.out");
_arguments = new List <object>();
_arguments.Add(_info.Corpus);
_arguments.Add(_info.Vocab);
_arguments.Add(_info.Edges);
_arguments.Add(_info.Links);
_arguments.Add(_outputFile);
_arguments.Add(_config.NumTopics);
_arguments.Add(_config.NumIterations);
_arguments.Add(_config.Alpha);
_arguments.Add(_config.Eta);
_arguments.Add(_config.RTMBeta);
_arguments.Add(_config.PredictionBeta);
_arguments.Add(_config.Seed);
}
/// <summary>
/// Creates a table of results for input into PCA
/// </summary>
/// <param name="matrices">Array of matrices</param>
/// <returns>Table in R format</returns>
private string CreateTable(params TLSimilarityMatrix[] matrices)
{
if (matrices.Length < 2)
{
throw new RDataException("Must have at least 2 matrices.");
}
FileStream tableFile = RUtil.CreateCacheFile("PCA.table");
TextWriter tableWriter = new StreamWriter(tableFile);
tableWriter.Write("M1");
for (int i = 1; i < matrices.Length; i++)
{
if (matrices[i].Count != matrices[0].Count)
{
throw new RDataException("Matrices have different count of links.");
}
tableWriter.Write(String.Format("\tM{0}", i + 1));
}
tableWriter.WriteLine();
foreach (TLSingleLink link in matrices[0].AllLinks)
{
tableWriter.Write(String.Format("{0}_{1}\t{2}",
link.SourceArtifactId,
link.TargetArtifactId,
link.Score
));
for (int i = 1; i < matrices.Length; i++)
{
tableWriter.Write(String.Format("\t{0}", matrices[i].GetScoreForLink(link.SourceArtifactId, link.TargetArtifactId)));
}
tableWriter.WriteLine();
}
tableWriter.Flush();
tableWriter.Close();
return(tableFile.Name);
}
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);
}
