private Hashtable[] collectAllTerms(Cluster[] clusters)
{
Hashtable[] clusterTerms = new Hashtable[clusters.Length]; //collect all term counts for each cluster
for (int i = 0; i < clusterTerms.Length; i++)
{
clusterTerms[i] = new Hashtable();
IDictionaryEnumerator en = clusters[i].DocIds.GetEnumerator();
while (en.MoveNext())
{
short docId = Convert.ToInt16(en.Key);
DocTermItem[] docTerms = GetIndex.DocTerms(docId); //get all terms+counts for current doc
foreach (DocTermItem dt in docTerms)
{
int termId = dt.TermId;
short termCount = dt.TermCount;
TermCountItem tc = new TermCountItem(termId, termCount);
if (!clusterTerms[i].Contains(termId))
{
clusterTerms[i].Add(termId, tc);
}
else
{
TermCountItem existingTC = (TermCountItem)clusterTerms[i][termId];
existingTC.termCount = existingTC.termCount + termCount;
}
}
}
}
return(clusterTerms);
}
public int CompareTo(object o)
{
TermCountItem x = (TermCountItem)o;
if (x.termCount < termCount)
{
return(-1);
}
else if (x.termCount > termCount)
{
return(1);
}
else
{
return(0);
}
}
private TermCountItem[][] getSortedTermArrays(Hashtable[] clusterTerms, int clusterCount)
{
TermCountItem[][] sortedTerms = new TermCountItem[clusterCount][];
for (int i = 0; i < clusterCount; i++)
{
int cursor = 0;
sortedTerms[i] = new TermCountItem[clusterTerms[i].Count]; //array length = # of terms in cluster
IDictionaryEnumerator en = clusterTerms[i].GetEnumerator();
while (en.MoveNext())
{
sortedTerms[i][cursor++] = (TermCountItem)en.Value;
}
Array.Sort(sortedTerms[i]);
}
return(sortedTerms);
}