/// <summary>
/// Makes a subset of input dataset, containing only documents matching assigned IDS
/// </summary>
/// <param name="input">The input.</param>
/// <param name="assignedIDs">The assigned i ds.</param>
/// <param name="inverse">if set to <c>true</c> if will keep only given assigned IDS</param>
public void FilterSpaceModel(List <String> assignedIDs, ILogBuilder log, Boolean applyToUnknownCategory = false)
{
foreach (KeyValuePair <string, List <string> > pair in LabelToDocumentLinks.GetAllRelationShipByName())
{
List <SpaceDocumentModel> toRemoveDocuments = new List <SpaceDocumentModel>();
List <SpaceDocumentModel> toScanDocuments = new List <SpaceDocumentModel>();
Boolean doScan = true;
if (pair.Key == SpaceLabel.UNKNOWN)
{
doScan = applyToUnknownCategory;
}
if (doScan)
{
toScanDocuments = LabelToDocumentLinks.GetAllLinkedB(pair.Key);
foreach (var vec in toScanDocuments)
{
if (!assignedIDs.Contains(vec.name))
{
toRemoveDocuments.Add(vec);
}
}
foreach (var vec in toRemoveDocuments)
{
documents.Remove(vec);
LabelToDocumentLinks.Remove(vec);
}
Double removal = toRemoveDocuments.Count.GetRatio(toScanDocuments.Count);
log.log("Vector count of [" + pair.Key + "] reduced to [" + removal.ToString("P2") + "] by document selection list");
}
}
}
public SpaceModel Clone()
{
SpaceModel output = new SpaceModel();
output.terms = terms.Clone();
output.terms_known_label = terms_known_label.Clone();
output.terms_unknown_label = terms_unknown_label.Clone();
output.documents = documents.Clone <SpaceDocumentModel>(false);
output.categories = categories.Clone <SpaceCategoryModel>(false);
output.topics = topics.CloneTerm <SpaceTopic>();
output.labels = labels.CloneTerm <SpaceLabel>();
foreach (var label in labels)
{
var linked = LabelToDocumentLinks.GetAllLinked(label);
var label2 = output.labels.First(x => x.name == label.name);
foreach (var l in linked)
{
SpaceDocumentModel doc2 = output.documents.First(x => x.name == l.name);
output.LabelToDocumentLinks.Add(label2, doc2, l.weight);
}
}
foreach (var label in labels)
{
var linked = LabelToCategoryLinks.GetAllLinked(label);
var label2 = output.labels.First(x => x.name == label.name);
foreach (var l in linked)
{
SpaceCategoryModel doc2 = output.categories.First(x => x.name == l.name);
output.LabelToCategoryLinks.Add(label2, doc2, l.weight);
}
}
return(output);
//output.categories
}
