private Set <int> ComputeCentroid(Cluster cluster)
{
if (cluster.Children.Count == 0)
{
ClusterInfo clusterInfo = new ClusterInfo();
clusterInfo.Centroid = ModelUtils.ComputeCentroid(cluster.Items, mDataset, CentroidType.Sum);
mModel.Add(cluster, clusterInfo);
return(cluster.Items);
}
else
{
Set <int> items = new Set <int>();
foreach (Cluster child in cluster.Children)
{
items.AddRange(ComputeCentroid(child));
}
if (mRemoveDuplicates)
{
items = Set <int> .Difference(cluster.Items, items);
}
ClusterInfo clusterInfo = new ClusterInfo();
clusterInfo.Centroid = ModelUtils.ComputeCentroid(items, mDataset, CentroidType.Sum);
mModel.Add(cluster, clusterInfo);
return(items);
}
}
public void ComputeFeatures()
{
foreach (Text text in mTexts)
{
text.ComputeFeatures();
foreach (KeyValuePair <string, double> feature in text.mFeatures)
{
AddFeatureVal(feature.Key, feature.Value);
}
}
foreach (string fvName in mTexts[0].mFeatureVectors.Keys)
{
ArrayList <SparseVector <double> > tmp = new ArrayList <SparseVector <double> >();
foreach (Text text in mTexts)
{
tmp.Add(text.mFeatureVectors[fvName]);
}
mFeatureVectors.Add(fvName, ModelUtils.ComputeCentroid(tmp, CentroidType.NrmL2));
}
}
